Task
stringclasses 9
values | Dataset
stringclasses 35
values | subset
stringclasses 1
value | Samples
listlengths 0
0
| Instance
dict | instruction
stringclasses 9
values | response
stringlengths 1
4.68k
| option
stringclasses 61
values | label_json
stringlengths 14
7.45k
| grouth_truth
stringlengths 1
4.68k
| instruction_v3
stringclasses 9
values | prompt
listlengths 2
2
| answer
stringlengths 14
7.45k
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( discover vulnerability : noticed ; the Plone 0day exploit : vulnerability ; he : discoverer ; working against the CMS backend : capabilities )",
"id": "3680",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( discover vulnerability : noticed ; the Plone 0day exploit : vulnerability ; he : discoverer ; working against the CMS backend : capabilities )",
"output_json": "{\"events\": [{\"event_type\": \"discover vulnerability\", \"trigger\": \"noticed\", \"arguments\": [{\"name\": \"the Plone 0day exploit\", \"role\": \"vulnerability\"}, {\"name\": \"he\", \"role\": \"discoverer\"}, {\"name\": \"working against the CMS backend\", \"role\": \"capabilities\"}]}]}",
"sentence": "The expert added further info on the attack , while experts at the FBI were working to fix the issue , he noticed that the Plone 0day exploit was still working against the CMS backend . )"
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( discover vulnerability : noticed ; the Plone 0day exploit : vulnerability ; he : discoverer ; working against the CMS backend : capabilities )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "discover vulnerability", "trigger": "noticed", "arguments": [{"name": "the Plone 0day exploit", "role": "vulnerability"}, {"name": "he", "role": "discoverer"}, {"name": "working against the CMS backend", "role": "capabilities"}]}]}
|
( discover vulnerability : noticed ; the Plone 0day exploit : vulnerability ; he : discoverer ; working against the CMS backend : capabilities )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nThe expert added further info on the attack , while experts at the FBI were working to fix the issue , he noticed that the Plone 0day exploit was still working against the CMS backend . )\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "discover vulnerability", "trigger": "noticed", "arguments": [{"name": "the Plone 0day exploit", "role": "vulnerability"}, {"name": "he", "role": "discoverer"}, {"name": "working against the CMS backend", "role": "capabilities"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( patch vulnerability : were running ; 2007 : time ; they : releaser ; ver 6.2 - RELEASE : patch number ; FreeBSD : vulnerable system )",
"id": "3681",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( patch vulnerability : were running ; 2007 : time ; they : releaser ; ver 6.2 - RELEASE : patch number ; FreeBSD : vulnerable system )",
"output_json": "{\"events\": [{\"event_type\": \"patch vulnerability\", \"trigger\": \"were running\", \"arguments\": [{\"name\": \"2007\", \"role\": \"time\"}, {\"name\": \"they\", \"role\": \"releaser\"}, {\"name\": \"ver 6.2 - RELEASE\", \"role\": \"patch number\"}, {\"name\": \"FreeBSD\", \"role\": \"vulnerable system\"}]}]}",
"sentence": ", but I was able to recon that they were running FreeBSD ver 6.2 - RELEASE that dates back to 2007 with their own custom configurations ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( patch vulnerability : were running ; 2007 : time ; they : releaser ; ver 6.2 - RELEASE : patch number ; FreeBSD : vulnerable system )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "patch vulnerability", "trigger": "were running", "arguments": [{"name": "2007", "role": "time"}, {"name": "they", "role": "releaser"}, {"name": "ver 6.2 - RELEASE", "role": "patch number"}, {"name": "FreeBSD", "role": "vulnerable system"}]}]}
|
( patch vulnerability : were running ; 2007 : time ; they : releaser ; ver 6.2 - RELEASE : patch number ; FreeBSD : vulnerable system )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\n, but I was able to recon that they were running FreeBSD ver 6.2 - RELEASE that dates back to 2007 with their own custom configurations .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "patch vulnerability", "trigger": "were running", "arguments": [{"name": "2007", "role": "time"}, {"name": "they", "role": "releaser"}, {"name": "ver 6.2 - RELEASE", "role": "patch number"}, {"name": "FreeBSD", "role": "vulnerable system"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( discover vulnerability : tweeted out ; FBI : vulnerable system owner ; I : discoverer ; my findings : vulnerability )",
"id": "3682",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( discover vulnerability : tweeted out ; FBI : vulnerable system owner ; I : discoverer ; my findings : vulnerability )",
"output_json": "{\"events\": [{\"event_type\": \"discover vulnerability\", \"trigger\": \"tweeted out\", \"arguments\": [{\"name\": \"FBI\", \"role\": \"vulnerable system owner\"}, {\"name\": \"I\", \"role\": \"discoverer\"}, {\"name\": \"my findings\", \"role\": \"vulnerability\"}]}]}",
"sentence": ", but still I did n’t leak out the whole contents of the backup files , instead I tweeted out my findings and thought to wait for FBI ’s response \" Now let ’s sit and wait for the FBI ’s response ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( discover vulnerability : tweeted out ; FBI : vulnerable system owner ; I : discoverer ; my findings : vulnerability )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "discover vulnerability", "trigger": "tweeted out", "arguments": [{"name": "FBI", "role": "vulnerable system owner"}, {"name": "I", "role": "discoverer"}, {"name": "my findings", "role": "vulnerability"}]}]}
|
( discover vulnerability : tweeted out ; FBI : vulnerable system owner ; I : discoverer ; my findings : vulnerability )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\n, but still I did n’t leak out the whole contents of the backup files , instead I tweeted out my findings and thought to wait for FBI ’s response \" Now let ’s sit and wait for the FBI ’s response .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "discover vulnerability", "trigger": "tweeted out", "arguments": [{"name": "FBI", "role": "vulnerable system owner"}, {"name": "I", "role": "discoverer"}, {"name": "my findings", "role": "vulnerability"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( discover vulnerability : publish ; 0day : vulnerability ; I : discoverer )",
"id": "3683",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( discover vulnerability : publish ; 0day : vulnerability ; I : discoverer )",
"output_json": "{\"events\": [{\"event_type\": \"discover vulnerability\", \"trigger\": \"publish\", \"arguments\": [{\"name\": \"0day\", \"role\": \"vulnerability\"}, {\"name\": \"I\", \"role\": \"discoverer\"}]}]}",
"sentence": "I obviously can not publish the 0day attack vector myself ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( discover vulnerability : publish ; 0day : vulnerability ; I : discoverer )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "discover vulnerability", "trigger": "publish", "arguments": [{"name": "0day", "role": "vulnerability"}, {"name": "I", "role": "discoverer"}]}]}
|
( discover vulnerability : publish ; 0day : vulnerability ; I : discoverer )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nI obviously can not publish the 0day attack vector myself .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "discover vulnerability", "trigger": "publish", "arguments": [{"name": "0day", "role": "vulnerability"}, {"name": "I", "role": "discoverer"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( discover vulnerability : confirmed ; the 0 - day : vulnerability ; The hacker : discoverer ; offered for sale : capabilities )",
"id": "3684",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( discover vulnerability : confirmed ; the 0 - day : vulnerability ; The hacker : discoverer ; offered for sale : capabilities )",
"output_json": "{\"events\": [{\"event_type\": \"discover vulnerability\", \"trigger\": \"confirmed\", \"arguments\": [{\"name\": \"the 0 - day\", \"role\": \"vulnerability\"}, {\"name\": \"The hacker\", \"role\": \"discoverer\"}, {\"name\": \"offered for sale\", \"role\": \"capabilities\"}]}]}",
"sentence": "The hacker confirmed that the 0 - day is offered for sale on Tor by a hacker that goes by the moniker “ lo4fer ” ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( discover vulnerability : confirmed ; the 0 - day : vulnerability ; The hacker : discoverer ; offered for sale : capabilities )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "discover vulnerability", "trigger": "confirmed", "arguments": [{"name": "the 0 - day", "role": "vulnerability"}, {"name": "The hacker", "role": "discoverer"}, {"name": "offered for sale", "role": "capabilities"}]}]}
|
( discover vulnerability : confirmed ; the 0 - day : vulnerability ; The hacker : discoverer ; offered for sale : capabilities )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nThe hacker confirmed that the 0 - day is offered for sale on Tor by a hacker that goes by the moniker “ lo4fer ” .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "discover vulnerability", "trigger": "confirmed", "arguments": [{"name": "the 0 - day", "role": "vulnerability"}, {"name": "The hacker", "role": "discoverer"}, {"name": "offered for sale", "role": "capabilities"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( discover vulnerability : tweet out ; the Plone CMS 0day attack vector : vulnerability ; I : discoverer )",
"id": "3685",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( discover vulnerability : tweet out ; the Plone CMS 0day attack vector : vulnerability ; I : discoverer )",
"output_json": "{\"events\": [{\"event_type\": \"discover vulnerability\", \"trigger\": \"tweet out\", \"arguments\": [{\"name\": \"the Plone CMS 0day attack vector\", \"role\": \"vulnerability\"}, {\"name\": \"I\", \"role\": \"discoverer\"}]}]}",
"sentence": "Once this 0day is no longer being sold , I will tweet out the Plone CMS 0day attack vector myself ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( discover vulnerability : tweet out ; the Plone CMS 0day attack vector : vulnerability ; I : discoverer )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "discover vulnerability", "trigger": "tweet out", "arguments": [{"name": "the Plone CMS 0day attack vector", "role": "vulnerability"}, {"name": "I", "role": "discoverer"}]}]}
|
( discover vulnerability : tweet out ; the Plone CMS 0day attack vector : vulnerability ; I : discoverer )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nOnce this 0day is no longer being sold , I will tweet out the Plone CMS 0day attack vector myself .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "discover vulnerability", "trigger": "tweet out", "arguments": [{"name": "the Plone CMS 0day attack vector", "role": "vulnerability"}, {"name": "I", "role": "discoverer"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( discover vulnerability : claims ; A miscreant : discoverer ; a zero day flaw : vulnerability ; FBI.gov : vulnerable system ; infiltrated Plone CMS : capabilities ; @cyberzeist : discoverer )",
"id": "3686",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( discover vulnerability : claims ; A miscreant : discoverer ; a zero day flaw : vulnerability ; FBI.gov : vulnerable system ; infiltrated Plone CMS : capabilities ; @cyberzeist : discoverer )",
"output_json": "{\"events\": [{\"event_type\": \"discover vulnerability\", \"trigger\": \"claims\", \"arguments\": [{\"name\": \"A miscreant\", \"role\": \"discoverer\"}, {\"name\": \"a zero day flaw\", \"role\": \"vulnerability\"}, {\"name\": \"FBI.gov\", \"role\": \"vulnerable system\"}, {\"name\": \"infiltrated Plone CMS\", \"role\": \"capabilities\"}, {\"name\": \"@cyberzeist\", \"role\": \"discoverer\"}]}]}",
"sentence": "A miscreant using the handle @cyberzeist claims to have infiltrated Plone CMS used by FBI.gov , using a zero day flaw allegedly for sale on an unnamed dark web site ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( discover vulnerability : claims ; A miscreant : discoverer ; a zero day flaw : vulnerability ; FBI.gov : vulnerable system ; infiltrated Plone CMS : capabilities ; @cyberzeist : discoverer )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "discover vulnerability", "trigger": "claims", "arguments": [{"name": "A miscreant", "role": "discoverer"}, {"name": "a zero day flaw", "role": "vulnerability"}, {"name": "FBI.gov", "role": "vulnerable system"}, {"name": "infiltrated Plone CMS", "role": "capabilities"}, {"name": "@cyberzeist", "role": "discoverer"}]}]}
|
( discover vulnerability : claims ; A miscreant : discoverer ; a zero day flaw : vulnerability ; FBI.gov : vulnerable system ; infiltrated Plone CMS : capabilities ; @cyberzeist : discoverer )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nA miscreant using the handle @cyberzeist claims to have infiltrated Plone CMS used by FBI.gov , using a zero day flaw allegedly for sale on an unnamed dark web site .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "discover vulnerability", "trigger": "claims", "arguments": [{"name": "A miscreant", "role": "discoverer"}, {"name": "a zero day flaw", "role": "vulnerability"}, {"name": "FBI.gov", "role": "vulnerable system"}, {"name": "infiltrated Plone CMS", "role": "capabilities"}, {"name": "@cyberzeist", "role": "discoverer"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( patch vulnerability : patching ; the vulnerability : vulnerability ; FBI : releaser ; permit public access : issues addressed )",
"id": "3687",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( patch vulnerability : patching ; the vulnerability : vulnerability ; FBI : releaser ; permit public access : issues addressed )",
"output_json": "{\"events\": [{\"event_type\": \"patch vulnerability\", \"trigger\": \"patching\", \"arguments\": [{\"name\": \"the vulnerability\", \"role\": \"vulnerability\"}, {\"name\": \"FBI\", \"role\": \"releaser\"}, {\"name\": \"permit public access\", \"role\": \"issues addressed\"}]}]}",
"sentence": "Cyberzeist claims to have conducted the hack last month and has posted to Twitter what they claim are screen captures showing the FBI patching against the vulnerability , which appeared to permit public access ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( patch vulnerability : patching ; the vulnerability : vulnerability ; FBI : releaser ; permit public access : issues addressed )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "patch vulnerability", "trigger": "patching", "arguments": [{"name": "the vulnerability", "role": "vulnerability"}, {"name": "FBI", "role": "releaser"}, {"name": "permit public access", "role": "issues addressed"}]}]}
|
( patch vulnerability : patching ; the vulnerability : vulnerability ; FBI : releaser ; permit public access : issues addressed )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nCyberzeist claims to have conducted the hack last month and has posted to Twitter what they claim are screen captures showing the FBI patching against the vulnerability , which appeared to permit public access .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "patch vulnerability", "trigger": "patching", "arguments": [{"name": "the vulnerability", "role": "vulnerability"}, {"name": "FBI", "role": "releaser"}, {"name": "permit public access", "role": "issues addressed"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( discover vulnerability : claiming ; vulnerability : vulnerability )",
"id": "3688",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( discover vulnerability : claiming ; vulnerability : vulnerability )",
"output_json": "{\"events\": [{\"event_type\": \"discover vulnerability\", \"trigger\": \"claiming\", \"arguments\": [{\"name\": \"vulnerability\", \"role\": \"vulnerability\"}]}]}",
"sentence": "The hacker dumped the 155 purported stolen credentials to online clipboard pastebin , claiming a vulnerability resides in a Plone Python module ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( discover vulnerability : claiming ; vulnerability : vulnerability )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "discover vulnerability", "trigger": "claiming", "arguments": [{"name": "vulnerability", "role": "vulnerability"}]}]}
|
( discover vulnerability : claiming ; vulnerability : vulnerability )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nThe hacker dumped the 155 purported stolen credentials to online clipboard pastebin , claiming a vulnerability resides in a Plone Python module .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "discover vulnerability", "trigger": "claiming", "arguments": [{"name": "vulnerability", "role": "vulnerability"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( phishing : pretending to come ; phishing scam email : tool ; Vodafone : trusted entity )",
"id": "3689",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( phishing : pretending to come ; phishing scam email : tool ; Vodafone : trusted entity )",
"output_json": "{\"events\": [{\"event_type\": \"phishing\", \"trigger\": \"pretending to come\", \"arguments\": [{\"name\": \"phishing scam email\", \"role\": \"tool\"}, {\"name\": \"Vodafone\", \"role\": \"trusted entity\"}]}]}",
"sentence": "ESET Ireland warns of an authentic looking phishing scam email , pretending to come from Vodafone ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( phishing : pretending to come ; phishing scam email : tool ; Vodafone : trusted entity )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "phishing", "trigger": "pretending to come", "arguments": [{"name": "phishing scam email", "role": "tool"}, {"name": "Vodafone", "role": "trusted entity"}]}]}
|
( phishing : pretending to come ; phishing scam email : tool ; Vodafone : trusted entity )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nESET Ireland warns of an authentic looking phishing scam email , pretending to come from Vodafone .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "phishing", "trigger": "pretending to come", "arguments": [{"name": "phishing scam email", "role": "tool"}, {"name": "Vodafone", "role": "trusted entity"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( phishing : pretending to be ; An email : tool ; Vodafone : trusted entity ; has been spamming Irish mailboxes : attack pattern )",
"id": "3690",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( phishing : pretending to be ; An email : tool ; Vodafone : trusted entity ; has been spamming Irish mailboxes : attack pattern )",
"output_json": "{\"events\": [{\"event_type\": \"phishing\", \"trigger\": \"pretending to be\", \"arguments\": [{\"name\": \"An email\", \"role\": \"tool\"}, {\"name\": \"Vodafone\", \"role\": \"trusted entity\"}, {\"name\": \"has been spamming Irish mailboxes\", \"role\": \"attack pattern\"}]}]}",
"sentence": "An email , pretending to be from Vodafone has been spamming Irish mailboxes with a phishing attempt ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( phishing : pretending to be ; An email : tool ; Vodafone : trusted entity ; has been spamming Irish mailboxes : attack pattern )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "phishing", "trigger": "pretending to be", "arguments": [{"name": "An email", "role": "tool"}, {"name": "Vodafone", "role": "trusted entity"}, {"name": "has been spamming Irish mailboxes", "role": "attack pattern"}]}]}
|
( phishing : pretending to be ; An email : tool ; Vodafone : trusted entity ; has been spamming Irish mailboxes : attack pattern )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nAn email , pretending to be from Vodafone has been spamming Irish mailboxes with a phishing attempt .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "phishing", "trigger": "pretending to be", "arguments": [{"name": "An email", "role": "tool"}, {"name": "Vodafone", "role": "trusted entity"}, {"name": "has been spamming Irish mailboxes", "role": "attack pattern"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( phishing : convincing ; the email : tool ; the Vodafone logos : trusted entity ; the email : tool ; a fake website : tool ; Mexico : place )",
"id": "3691",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( phishing : convincing ; the email : tool ; the Vodafone logos : trusted entity ; the email : tool ; a fake website : tool ; Mexico : place )",
"output_json": "{\"events\": [{\"event_type\": \"phishing\", \"trigger\": \"convincing\", \"arguments\": [{\"name\": \"the email\", \"role\": \"tool\"}, {\"name\": \"the Vodafone logos\", \"role\": \"trusted entity\"}, {\"name\": \"the email\", \"role\": \"tool\"}, {\"name\": \"a fake website\", \"role\": \"tool\"}, {\"name\": \"Mexico\", \"role\": \"place\"}]}]}",
"sentence": "While the email is made to look very convincing , with all the Vodafone logos and overall appearance , all the links in the email lead to a fake website , registered in Mexico , which tries to trick the user into submitting their account info and payment details ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( phishing : convincing ; the email : tool ; the Vodafone logos : trusted entity ; the email : tool ; a fake website : tool ; Mexico : place )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "phishing", "trigger": "convincing", "arguments": [{"name": "the email", "role": "tool"}, {"name": "the Vodafone logos", "role": "trusted entity"}, {"name": "the email", "role": "tool"}, {"name": "a fake website", "role": "tool"}, {"name": "Mexico", "role": "place"}]}]}
|
( phishing : convincing ; the email : tool ; the Vodafone logos : trusted entity ; the email : tool ; a fake website : tool ; Mexico : place )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nWhile the email is made to look very convincing , with all the Vodafone logos and overall appearance , all the links in the email lead to a fake website , registered in Mexico , which tries to trick the user into submitting their account info and payment details .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "phishing", "trigger": "convincing", "arguments": [{"name": "the email", "role": "tool"}, {"name": "the Vodafone logos", "role": "trusted entity"}, {"name": "the email", "role": "tool"}, {"name": "a fake website", "role": "tool"}, {"name": "Mexico", "role": "place"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : the breach ; over the winter break : time ; Valley Glen : place ; Los Angeles Valley College : victim )",
"id": "3692",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : the breach ; over the winter break : time ; Valley Glen : place ; Los Angeles Valley College : victim )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"the breach\", \"arguments\": [{\"name\": \"over the winter break\", \"role\": \"time\"}, {\"name\": \"Valley Glen\", \"role\": \"place\"}, {\"name\": \"Los Angeles Valley College\", \"role\": \"victim\"}]}]}",
"sentence": "Los Angeles Valley College in Valley Glen was subject to a cyber attack over the winter break but it is not known how large the breach was , officials said Tuesday ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : the breach ; over the winter break : time ; Valley Glen : place ; Los Angeles Valley College : victim )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "the breach", "arguments": [{"name": "over the winter break", "role": "time"}, {"name": "Valley Glen", "role": "place"}, {"name": "Los Angeles Valley College", "role": "victim"}]}]}
|
( data breach : the breach ; over the winter break : time ; Valley Glen : place ; Los Angeles Valley College : victim )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nLos Angeles Valley College in Valley Glen was subject to a cyber attack over the winter break but it is not known how large the breach was , officials said Tuesday .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "the breach", "arguments": [{"name": "over the winter break", "role": "time"}, {"name": "Valley Glen", "role": "place"}, {"name": "Los Angeles Valley College", "role": "victim"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : was compromised ; personal data : compromised data ; anyone : victim )",
"id": "3693",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : was compromised ; personal data : compromised data ; anyone : victim )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"was compromised\", \"arguments\": [{\"name\": \"personal data\", \"role\": \"compromised data\"}, {\"name\": \"anyone\", \"role\": \"victim\"}]}]}",
"sentence": "Additional details about the attack were not made available and it was not immediately clear if anyone ’s personal data was compromised ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : was compromised ; personal data : compromised data ; anyone : victim )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "was compromised", "arguments": [{"name": "personal data", "role": "compromised data"}, {"name": "anyone", "role": "victim"}]}]}
|
( data breach : was compromised ; personal data : compromised data ; anyone : victim )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nAdditional details about the attack were not made available and it was not immediately clear if anyone ’s personal data was compromised .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "was compromised", "arguments": [{"name": "personal data", "role": "compromised data"}, {"name": "anyone", "role": "victim"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( discover vulnerability : uncovered ; in December of 2015 and June of 2016 : time ; vulnerabilities : vulnerability ; Topps databases : vulnerable system ; security researcher Chris Vickery : discoverer )",
"id": "3694",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( discover vulnerability : uncovered ; in December of 2015 and June of 2016 : time ; vulnerabilities : vulnerability ; Topps databases : vulnerable system ; security researcher Chris Vickery : discoverer )",
"output_json": "{\"events\": [{\"event_type\": \"discover vulnerability\", \"trigger\": \"uncovered\", \"arguments\": [{\"name\": \"in December of 2015 and June of 2016\", \"role\": \"time\"}, {\"name\": \"vulnerabilities\", \"role\": \"vulnerability\"}, {\"name\": \"Topps databases\", \"role\": \"vulnerable system\"}, {\"name\": \"security researcher Chris Vickery\", \"role\": \"discoverer\"}]}]}",
"sentence": "BBC News notes that security researcher Chris Vickery uncovered vulnerabilities in Topps databases back in December of 2015 and June of 2016 , but he was n't able to get a response from the company by email ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( discover vulnerability : uncovered ; in December of 2015 and June of 2016 : time ; vulnerabilities : vulnerability ; Topps databases : vulnerable system ; security researcher Chris Vickery : discoverer )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "discover vulnerability", "trigger": "uncovered", "arguments": [{"name": "in December of 2015 and June of 2016", "role": "time"}, {"name": "vulnerabilities", "role": "vulnerability"}, {"name": "Topps databases", "role": "vulnerable system"}, {"name": "security researcher Chris Vickery", "role": "discoverer"}]}]}
|
( discover vulnerability : uncovered ; in December of 2015 and June of 2016 : time ; vulnerabilities : vulnerability ; Topps databases : vulnerable system ; security researcher Chris Vickery : discoverer )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nBBC News notes that security researcher Chris Vickery uncovered vulnerabilities in Topps databases back in December of 2015 and June of 2016 , but he was n't able to get a response from the company by email .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "discover vulnerability", "trigger": "uncovered", "arguments": [{"name": "in December of 2015 and June of 2016", "role": "time"}, {"name": "vulnerabilities", "role": "vulnerability"}, {"name": "Topps databases", "role": "vulnerable system"}, {"name": "security researcher Chris Vickery", "role": "discoverer"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( discover vulnerability : found ; Vickery : discoverer ; flaws : vulnerability )",
"id": "3695",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( discover vulnerability : found ; Vickery : discoverer ; flaws : vulnerability )",
"output_json": "{\"events\": [{\"event_type\": \"discover vulnerability\", \"trigger\": \"found\", \"arguments\": [{\"name\": \"Vickery\", \"role\": \"discoverer\"}, {\"name\": \"flaws\", \"role\": \"vulnerability\"}]}]}",
"sentence": "It 's not clear whether the credit card breach was related to the flaws Vickery found ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( discover vulnerability : found ; Vickery : discoverer ; flaws : vulnerability )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "discover vulnerability", "trigger": "found", "arguments": [{"name": "Vickery", "role": "discoverer"}, {"name": "flaws", "role": "vulnerability"}]}]}
|
( discover vulnerability : found ; Vickery : discoverer ; flaws : vulnerability )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nIt 's not clear whether the credit card breach was related to the flaws Vickery found .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "discover vulnerability", "trigger": "found", "arguments": [{"name": "Vickery", "role": "discoverer"}, {"name": "flaws", "role": "vulnerability"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : were stolen ; password : compromised data ; user name : compromised data ; they : victim )",
"id": "3696",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : were stolen ; password : compromised data ; user name : compromised data ; they : victim )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"were stolen\", \"arguments\": [{\"name\": \"password\", \"role\": \"compromised data\"}, {\"name\": \"user name\", \"role\": \"compromised data\"}, {\"name\": \"they\", \"role\": \"victim\"}]}]}",
"sentence": "Seventy percent said they would stop using digital payments if money was stolen from a linked bank account , 68 percent said they would do so if unauthorized charges appeared on a linked credit card account , and 59 percent said they would do so if their user name and password were stolen ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : were stolen ; password : compromised data ; user name : compromised data ; they : victim )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "were stolen", "arguments": [{"name": "password", "role": "compromised data"}, {"name": "user name", "role": "compromised data"}, {"name": "they", "role": "victim"}]}]}
|
( data breach : were stolen ; password : compromised data ; user name : compromised data ; they : victim )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nSeventy percent said they would stop using digital payments if money was stolen from a linked bank account , 68 percent said they would do so if unauthorized charges appeared on a linked credit card account , and 59 percent said they would do so if their user name and password were stolen .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "were stolen", "arguments": [{"name": "password", "role": "compromised data"}, {"name": "user name", "role": "compromised data"}, {"name": "they", "role": "victim"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : been stolen ; its systems : victim ; credit card data : compromised data )",
"id": "3697",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : been stolen ; its systems : victim ; credit card data : compromised data )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"been stolen\", \"arguments\": [{\"name\": \"its systems\", \"role\": \"victim\"}, {\"name\": \"credit card data\", \"role\": \"compromised data\"}]}]}",
"sentence": "A separate Thales survey of 1,000 adults in the U.S. and the U.K. recently found that 55 percent of respondents would switch to only using cash at a retailer if they learned that credit card data had been stolen from its systems -- and 20 percent would stop shopping at that retailer altogether"
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : been stolen ; its systems : victim ; credit card data : compromised data )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "been stolen", "arguments": [{"name": "its systems", "role": "victim"}, {"name": "credit card data", "role": "compromised data"}]}]}
|
( data breach : been stolen ; its systems : victim ; credit card data : compromised data )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nA separate Thales survey of 1,000 adults in the U.S. and the U.K. recently found that 55 percent of respondents would switch to only using cash at a retailer if they learned that credit card data had been stolen from its systems -- and 20 percent would stop shopping at that retailer altogether\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "been stolen", "arguments": [{"name": "its systems", "role": "victim"}, {"name": "credit card data", "role": "compromised data"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : stole ; he : attacker )",
"id": "3698",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : stole ; he : attacker )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"stole\", \"arguments\": [{\"name\": \"he\", \"role\": \"attacker\"}]}]}",
"sentence": "The attacker also contacted the KeepKey staff , offering to provide details about how he hacked the Verizon email account and what he stole ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : stole ; he : attacker )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "stole", "arguments": [{"name": "he", "role": "attacker"}]}]}
|
( data breach : stole ; he : attacker )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nThe attacker also contacted the KeepKey staff , offering to provide details about how he hacked the Verizon email account and what he stole .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "stole", "arguments": [{"name": "he", "role": "attacker"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : the hack ; the stolen data : compromised data ; The attacker : attacker )",
"id": "3699",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : the hack ; the stolen data : compromised data ; The attacker : attacker )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"the hack\", \"arguments\": [{\"name\": \"the stolen data\", \"role\": \"compromised data\"}, {\"name\": \"The attacker\", \"role\": \"attacker\"}]}]}",
"sentence": "The attacker had also promised to return the stolen data and keep quiet about the hack if KeepKey would agree to pay him 30 Bitcoin ( ~ $ 30,000 ) ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : the hack ; the stolen data : compromised data ; The attacker : attacker )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "the hack", "arguments": [{"name": "the stolen data", "role": "compromised data"}, {"name": "The attacker", "role": "attacker"}]}]}
|
( data breach : the hack ; the stolen data : compromised data ; The attacker : attacker )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nThe attacker had also promised to return the stolen data and keep quiet about the hack if KeepKey would agree to pay him 30 Bitcoin ( ~ $ 30,000 ) .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "the hack", "arguments": [{"name": "the stolen data", "role": "compromised data"}, {"name": "The attacker", "role": "attacker"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( ransom : paying the ransom demand ; the KeepKey team : victim ; the attacker : attacker )",
"id": "3700",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( ransom : paying the ransom demand ; the KeepKey team : victim ; the attacker : attacker )",
"output_json": "{\"events\": [{\"event_type\": \"ransom\", \"trigger\": \"paying the ransom demand\", \"arguments\": [{\"name\": \"the KeepKey team\", \"role\": \"victim\"}, {\"name\": \"the attacker\", \"role\": \"attacker\"}]}]}",
"sentence": "Instead of paying the ransom demand , the KeepKey team managed to stall the attacker for two more hours , during which time they regained access to all but one account , the company 's Twitter profile ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( ransom : paying the ransom demand ; the KeepKey team : victim ; the attacker : attacker )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "ransom", "trigger": "paying the ransom demand", "arguments": [{"name": "the KeepKey team", "role": "victim"}, {"name": "the attacker", "role": "attacker"}]}]}
|
( ransom : paying the ransom demand ; the KeepKey team : victim ; the attacker : attacker )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nInstead of paying the ransom demand , the KeepKey team managed to stall the attacker for two more hours , during which time they regained access to all but one account , the company 's Twitter profile .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "ransom", "trigger": "paying the ransom demand", "arguments": [{"name": "the KeepKey team", "role": "victim"}, {"name": "the attacker", "role": "attacker"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( ransom : ransom ; 30 Bitcoin : price ; the company : victim ; attacker : attacker )",
"id": "3701",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( ransom : ransom ; 30 Bitcoin : price ; the company : victim ; attacker : attacker )",
"output_json": "{\"events\": [{\"event_type\": \"ransom\", \"trigger\": \"ransom\", \"arguments\": [{\"name\": \"30 Bitcoin\", \"role\": \"price\"}, {\"name\": \"the company\", \"role\": \"victim\"}, {\"name\": \"attacker\", \"role\": \"attacker\"}]}]}",
"sentence": "Since the night of the hack , the company has filed a complaint with the FBI and is now offering the 30 Bitcoin ransom as a reward for any clues that lead to the attacker 's arrest ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( ransom : ransom ; 30 Bitcoin : price ; the company : victim ; attacker : attacker )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "ransom", "trigger": "ransom", "arguments": [{"name": "30 Bitcoin", "role": "price"}, {"name": "the company", "role": "victim"}, {"name": "attacker", "role": "attacker"}]}]}
|
( ransom : ransom ; 30 Bitcoin : price ; the company : victim ; attacker : attacker )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nSince the night of the hack , the company has filed a complaint with the FBI and is now offering the 30 Bitcoin ransom as a reward for any clues that lead to the attacker 's arrest .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "ransom", "trigger": "ransom", "arguments": [{"name": "30 Bitcoin", "role": "price"}, {"name": "the company", "role": "victim"}, {"name": "attacker", "role": "attacker"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : steal ; home addresses : compromised data ; emails : compromised data ; phone numbers : compromised data ; the attacker : attacker ; users : victim )",
"id": "3702",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : steal ; home addresses : compromised data ; emails : compromised data ; phone numbers : compromised data ; the attacker : attacker ; users : victim )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"steal\", \"arguments\": [{\"name\": \"home addresses\", \"role\": \"compromised data\"}, {\"name\": \"emails\", \"role\": \"compromised data\"}, {\"name\": \"phone numbers\", \"role\": \"compromised data\"}, {\"name\": \"the attacker\", \"role\": \"attacker\"}, {\"name\": \"users\", \"role\": \"victim\"}]}]}",
"sentence": "In the Christmas security breach , the attacker would have only managed to steal home addresses , emails , and phone numbers from users that have bought KeepKey devices in the past , and not the content of those devices ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : steal ; home addresses : compromised data ; emails : compromised data ; phone numbers : compromised data ; the attacker : attacker ; users : victim )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "steal", "arguments": [{"name": "home addresses", "role": "compromised data"}, {"name": "emails", "role": "compromised data"}, {"name": "phone numbers", "role": "compromised data"}, {"name": "the attacker", "role": "attacker"}, {"name": "users", "role": "victim"}]}]}
|
( data breach : steal ; home addresses : compromised data ; emails : compromised data ; phone numbers : compromised data ; the attacker : attacker ; users : victim )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nIn the Christmas security breach , the attacker would have only managed to steal home addresses , emails , and phone numbers from users that have bought KeepKey devices in the past , and not the content of those devices .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "steal", "arguments": [{"name": "home addresses", "role": "compromised data"}, {"name": "emails", "role": "compromised data"}, {"name": "phone numbers", "role": "compromised data"}, {"name": "the attacker", "role": "attacker"}, {"name": "users", "role": "victim"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : steal ; attacker : attacker ; used the access over these accounts : attack pattern ; KeepKey customer : victim ; data : compromised data )",
"id": "3703",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : steal ; attacker : attacker ; used the access over these accounts : attack pattern ; KeepKey customer : victim ; data : compromised data )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"steal\", \"arguments\": [{\"name\": \"attacker\", \"role\": \"attacker\"}, {\"name\": \"used the access over these accounts\", \"role\": \"attack pattern\"}, {\"name\": \"KeepKey customer\", \"role\": \"victim\"}, {\"name\": \"data\", \"role\": \"compromised data\"}]}]}",
"sentence": "It is unknown at the time of writing if the attacker used the access over these accounts to steal any KeepKey customer data ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : steal ; attacker : attacker ; used the access over these accounts : attack pattern ; KeepKey customer : victim ; data : compromised data )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "steal", "arguments": [{"name": "attacker", "role": "attacker"}, {"name": "used the access over these accounts", "role": "attack pattern"}, {"name": "KeepKey customer", "role": "victim"}, {"name": "data", "role": "compromised data"}]}]}
|
( data breach : steal ; attacker : attacker ; used the access over these accounts : attack pattern ; KeepKey customer : victim ; data : compromised data )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nIt is unknown at the time of writing if the attacker used the access over these accounts to steal any KeepKey customer data .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "steal", "arguments": [{"name": "attacker", "role": "attacker"}, {"name": "used the access over these accounts", "role": "attack pattern"}, {"name": "KeepKey customer", "role": "victim"}, {"name": "data", "role": "compromised data"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : gain access ; attacker : attacker ; the sales distribution channels : victim ; their details : compromised data ; all customers : victim ; email marketing software accounts : compromised data )",
"id": "3704",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : gain access ; attacker : attacker ; the sales distribution channels : victim ; their details : compromised data ; all customers : victim ; email marketing software accounts : compromised data )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"gain access\", \"arguments\": [{\"name\": \"attacker\", \"role\": \"attacker\"}, {\"name\": \"the sales distribution channels\", \"role\": \"victim\"}, {\"name\": \"their details\", \"role\": \"compromised data\"}, {\"name\": \"all customers\", \"role\": \"victim\"}, {\"name\": \"email marketing software accounts\", \"role\": \"compromised data\"}]}]}",
"sentence": "Nevertheless , as a precautionary measure , the company is offering a 30 - day refund policy to all customers that had their details stored in the sales distribution channels and email marketing software accounts that the attacker managed to gain access to ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : gain access ; attacker : attacker ; the sales distribution channels : victim ; their details : compromised data ; all customers : victim ; email marketing software accounts : compromised data )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "gain access", "arguments": [{"name": "attacker", "role": "attacker"}, {"name": "the sales distribution channels", "role": "victim"}, {"name": "their details", "role": "compromised data"}, {"name": "all customers", "role": "victim"}, {"name": "email marketing software accounts", "role": "compromised data"}]}]}
|
( data breach : gain access ; attacker : attacker ; the sales distribution channels : victim ; their details : compromised data ; all customers : victim ; email marketing software accounts : compromised data )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nNevertheless , as a precautionary measure , the company is offering a 30 - day refund policy to all customers that had their details stored in the sales distribution channels and email marketing software accounts that the attacker managed to gain access to .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "gain access", "arguments": [{"name": "attacker", "role": "attacker"}, {"name": "the sales distribution channels", "role": "victim"}, {"name": "their details", "role": "compromised data"}, {"name": "all customers", "role": "victim"}, {"name": "email marketing software accounts", "role": "compromised data"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : downloading a copy ; Ethereum forum : victim ; database backup : compromised data ; April 2016 : time ; hacker : attacker ; reset the passwords for various accounts : attack pattern ; took over a mobile number : attack pattern ; Two weeks later : time ; Ethereum Project : victim ; admins : victim )",
"id": "3705",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : downloading a copy ; Ethereum forum : victim ; database backup : compromised data ; April 2016 : time ; hacker : attacker ; reset the passwords for various accounts : attack pattern ; took over a mobile number : attack pattern ; Two weeks later : time ; Ethereum Project : victim ; admins : victim )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"downloading a copy\", \"arguments\": [{\"name\": \"Ethereum forum\", \"role\": \"victim\"}, {\"name\": \"database backup\", \"role\": \"compromised data\"}, {\"name\": \"April 2016\", \"role\": \"time\"}, {\"name\": \"hacker\", \"role\": \"attacker\"}, {\"name\": \"reset the passwords for various accounts\", \"role\": \"attack pattern\"}, {\"name\": \"took over a mobile number\", \"role\": \"attack pattern\"}, {\"name\": \"Two weeks later\", \"role\": \"time\"}, {\"name\": \"Ethereum Project\", \"role\": \"victim\"}, {\"name\": \"admins\", \"role\": \"victim\"}]}]}",
"sentence": "Two weeks later , the same hacker took over a mobile number for one of the Ethereum Project 's admins and used it to reset the passwords for various accounts , eventually downloading a copy of Ethereum forum database backup , dated to April 2016 ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : downloading a copy ; Ethereum forum : victim ; database backup : compromised data ; April 2016 : time ; hacker : attacker ; reset the passwords for various accounts : attack pattern ; took over a mobile number : attack pattern ; Two weeks later : time ; Ethereum Project : victim ; admins : victim )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "downloading a copy", "arguments": [{"name": "Ethereum forum", "role": "victim"}, {"name": "database backup", "role": "compromised data"}, {"name": "April 2016", "role": "time"}, {"name": "hacker", "role": "attacker"}, {"name": "reset the passwords for various accounts", "role": "attack pattern"}, {"name": "took over a mobile number", "role": "attack pattern"}, {"name": "Two weeks later", "role": "time"}, {"name": "Ethereum Project", "role": "victim"}, {"name": "admins", "role": "victim"}]}]}
|
( data breach : downloading a copy ; Ethereum forum : victim ; database backup : compromised data ; April 2016 : time ; hacker : attacker ; reset the passwords for various accounts : attack pattern ; took over a mobile number : attack pattern ; Two weeks later : time ; Ethereum Project : victim ; admins : victim )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nTwo weeks later , the same hacker took over a mobile number for one of the Ethereum Project 's admins and used it to reset the passwords for various accounts , eventually downloading a copy of Ethereum forum database backup , dated to April 2016 .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "downloading a copy", "arguments": [{"name": "Ethereum forum", "role": "victim"}, {"name": "database backup", "role": "compromised data"}, {"name": "April 2016", "role": "time"}, {"name": "hacker", "role": "attacker"}, {"name": "reset the passwords for various accounts", "role": "attack pattern"}, {"name": "took over a mobile number", "role": "attack pattern"}, {"name": "Two weeks later", "role": "time"}, {"name": "Ethereum Project", "role": "victim"}, {"name": "admins", "role": "victim"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( discover vulnerability : reports ; the latest iOS 10 : vulnerable system version )",
"id": "3706",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( discover vulnerability : reports ; the latest iOS 10 : vulnerable system version )",
"output_json": "{\"events\": [{\"event_type\": \"discover vulnerability\", \"trigger\": \"reports\", \"arguments\": [{\"name\": \"the latest iOS 10\", \"role\": \"vulnerable system version\"}]}]}",
"sentence": "According to reports , there is a nefarious text message that is being sent to iPhone users even on those phones that run on the latest iOS 10 ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( discover vulnerability : reports ; the latest iOS 10 : vulnerable system version )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "discover vulnerability", "trigger": "reports", "arguments": [{"name": "the latest iOS 10", "role": "vulnerable system version"}]}]}
|
( discover vulnerability : reports ; the latest iOS 10 : vulnerable system version )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nAccording to reports , there is a nefarious text message that is being sent to iPhone users even on those phones that run on the latest iOS 10 .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "discover vulnerability", "trigger": "reports", "arguments": [{"name": "the latest iOS 10", "role": "vulnerable system version"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : a deanonymization attack ; Tor users : victim )",
"id": "3707",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : a deanonymization attack ; Tor users : victim )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"a deanonymization attack\", \"arguments\": [{\"name\": \"Tor users\", \"role\": \"victim\"}]}]}",
"sentence": "Speaking at last week 's 33rd Chaos Communication Congress , Vasilios Mavroudis , one of the six researchers , detailed a deanonymization attack on Tor users that leaks their real IP and a few other details ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : a deanonymization attack ; Tor users : victim )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "a deanonymization attack", "arguments": [{"name": "Tor users", "role": "victim"}]}]}
|
( data breach : a deanonymization attack ; Tor users : victim )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nSpeaking at last week 's 33rd Chaos Communication Congress , Vasilios Mavroudis , one of the six researchers , detailed a deanonymization attack on Tor users that leaks their real IP and a few other details .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "a deanonymization attack", "arguments": [{"name": "Tor users", "role": "victim"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( phishing : The attack ; None )",
"id": "3708",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( phishing : The attack ; None )",
"output_json": "{\"events\": [{\"event_type\": \"phishing\", \"trigger\": \"The attack\", \"arguments\": []}]}",
"sentence": "The attack that the research team put together relies on tricking a Tor user into accessing a web page that contains ads that emit ultrasounds or accessing a page that contains hidden JavaScript code that forces the browser to emit the ultrasounds via the HTML5 Audio API ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( phishing : The attack ; None )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "phishing", "trigger": "The attack", "arguments": []}]}
|
( phishing : The attack ; None )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nThe attack that the research team put together relies on tricking a Tor user into accessing a web page that contains ads that emit ultrasounds or accessing a page that contains hidden JavaScript code that forces the browser to emit the ultrasounds via the HTML5 Audio API .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "phishing", "trigger": "The attack", "arguments": []}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : has intercepted ; researcher : attacker ; device MAC address : compromised data ; IMEI code : compromised data ; telephone number : compromised data ; geo - location coordinates : compromised data ; the phone : victim ; Android ID : compromised data ; real IP address : compromised data ; user : victim ; Mavroudis : attacker )",
"id": "3709",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : has intercepted ; researcher : attacker ; device MAC address : compromised data ; IMEI code : compromised data ; telephone number : compromised data ; geo - location coordinates : compromised data ; the phone : victim ; Android ID : compromised data ; real IP address : compromised data ; user : victim ; Mavroudis : attacker )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"has intercepted\", \"arguments\": [{\"name\": \"researcher\", \"role\": \"attacker\"}, {\"name\": \"device MAC address\", \"role\": \"compromised data\"}, {\"name\": \"IMEI code\", \"role\": \"compromised data\"}, {\"name\": \"telephone number\", \"role\": \"compromised data\"}, {\"name\": \"geo - location coordinates\", \"role\": \"compromised data\"}, {\"name\": \"the phone\", \"role\": \"victim\"}, {\"name\": \"Android ID\", \"role\": \"compromised data\"}, {\"name\": \"real IP address\", \"role\": \"compromised data\"}, {\"name\": \"user\", \"role\": \"victim\"}, {\"name\": \"Mavroudis\", \"role\": \"attacker\"}]}]}",
"sentence": "In tests carried out by Mavroudis , the researcher has intercepted some of the traffic these ultrasound beacons trigger on behalf of the phone , traffic which contains details such as the user 's real IP address , geo - location coordinates , telephone number , Android ID , IMEI code , and device MAC address ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : has intercepted ; researcher : attacker ; device MAC address : compromised data ; IMEI code : compromised data ; telephone number : compromised data ; geo - location coordinates : compromised data ; the phone : victim ; Android ID : compromised data ; real IP address : compromised data ; user : victim ; Mavroudis : attacker )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "has intercepted", "arguments": [{"name": "researcher", "role": "attacker"}, {"name": "device MAC address", "role": "compromised data"}, {"name": "IMEI code", "role": "compromised data"}, {"name": "telephone number", "role": "compromised data"}, {"name": "geo - location coordinates", "role": "compromised data"}, {"name": "the phone", "role": "victim"}, {"name": "Android ID", "role": "compromised data"}, {"name": "real IP address", "role": "compromised data"}, {"name": "user", "role": "victim"}, {"name": "Mavroudis", "role": "attacker"}]}]}
|
( data breach : has intercepted ; researcher : attacker ; device MAC address : compromised data ; IMEI code : compromised data ; telephone number : compromised data ; geo - location coordinates : compromised data ; the phone : victim ; Android ID : compromised data ; real IP address : compromised data ; user : victim ; Mavroudis : attacker )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nIn tests carried out by Mavroudis , the researcher has intercepted some of the traffic these ultrasound beacons trigger on behalf of the phone , traffic which contains details such as the user 's real IP address , geo - location coordinates , telephone number , Android ID , IMEI code , and device MAC address .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "has intercepted", "arguments": [{"name": "researcher", "role": "attacker"}, {"name": "device MAC address", "role": "compromised data"}, {"name": "IMEI code", "role": "compromised data"}, {"name": "telephone number", "role": "compromised data"}, {"name": "geo - location coordinates", "role": "compromised data"}, {"name": "the phone", "role": "victim"}, {"name": "Android ID", "role": "compromised data"}, {"name": "real IP address", "role": "compromised data"}, {"name": "user", "role": "victim"}, {"name": "Mavroudis", "role": "attacker"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( ransom : ask ; victims : victim )",
"id": "3710",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( ransom : ask ; victims : victim )",
"output_json": "{\"events\": [{\"event_type\": \"ransom\", \"trigger\": \"ask\", \"arguments\": [{\"name\": \"victims\", \"role\": \"victim\"}]}]}",
"sentence": "The messages usually impersonate some law enforcement authority and ask victims to pay fictitious fines to regain control ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( ransom : ask ; victims : victim )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "ransom", "trigger": "ask", "arguments": [{"name": "victims", "role": "victim"}]}]}
|
( ransom : ask ; victims : victim )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nThe messages usually impersonate some law enforcement authority and ask victims to pay fictitious fines to regain control .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "ransom", "trigger": "ask", "arguments": [{"name": "victims", "role": "victim"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( ransom : paying ; $ 500 : price ; click on the payment section : payment method )",
"id": "3711",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( ransom : paying ; $ 500 : price ; click on the payment section : payment method )",
"output_json": "{\"events\": [{\"event_type\": \"ransom\", \"trigger\": \"paying\", \"arguments\": [{\"name\": \"$ 500\", \"role\": \"price\"}, {\"name\": \"click on the payment section\", \"role\": \"payment method\"}]}]}",
"sentence": "The ransom amount itself was $ 500 although even paying that would have been difficult because there was no way to click on the payment section to find the instructions on how to do so ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( ransom : paying ; $ 500 : price ; click on the payment section : payment method )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "ransom", "trigger": "paying", "arguments": [{"name": "$ 500", "role": "price"}, {"name": "click on the payment section", "role": "payment method"}]}]}
|
( ransom : paying ; $ 500 : price ; click on the payment section : payment method )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nThe ransom amount itself was $ 500 although even paying that would have been difficult because there was no way to click on the payment section to find the instructions on how to do so .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "ransom", "trigger": "paying", "arguments": [{"name": "$ 500", "role": "price"}, {"name": "click on the payment section", "role": "payment method"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( discover vulnerability : has found ; Anti-Virus product : vulnerable system ; SSL proxy packaged : vulnerable system ; Project Zero : discoverer ; an SSL certificate collision : vulnerability ; Kaspersky : vulnerable system owner ; Google : discoverer )",
"id": "3712",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( discover vulnerability : has found ; Anti-Virus product : vulnerable system ; SSL proxy packaged : vulnerable system ; Project Zero : discoverer ; an SSL certificate collision : vulnerability ; Kaspersky : vulnerable system owner ; Google : discoverer )",
"output_json": "{\"events\": [{\"event_type\": \"discover vulnerability\", \"trigger\": \"has found\", \"arguments\": [{\"name\": \"Anti-Virus product\", \"role\": \"vulnerable system\"}, {\"name\": \"SSL proxy packaged\", \"role\": \"vulnerable system\"}, {\"name\": \"Project Zero\", \"role\": \"discoverer\"}, {\"name\": \"an SSL certificate collision\", \"role\": \"vulnerability\"}, {\"name\": \"Kaspersky\", \"role\": \"vulnerable system owner\"}, {\"name\": \"Google\", \"role\": \"discoverer\"}]}]}",
"sentence": "By Chris Duckett Google 's Project Zero has found that it was previously trivial to create an SSL certificate collision thanks to Kaspersky using only the first 32 bits of an MD5 hash in its SSL proxy packaged with its Anti-Virus product ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( discover vulnerability : has found ; Anti-Virus product : vulnerable system ; SSL proxy packaged : vulnerable system ; Project Zero : discoverer ; an SSL certificate collision : vulnerability ; Kaspersky : vulnerable system owner ; Google : discoverer )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "discover vulnerability", "trigger": "has found", "arguments": [{"name": "Anti-Virus product", "role": "vulnerable system"}, {"name": "SSL proxy packaged", "role": "vulnerable system"}, {"name": "Project Zero", "role": "discoverer"}, {"name": "an SSL certificate collision", "role": "vulnerability"}, {"name": "Kaspersky", "role": "vulnerable system owner"}, {"name": "Google", "role": "discoverer"}]}]}
|
( discover vulnerability : has found ; Anti-Virus product : vulnerable system ; SSL proxy packaged : vulnerable system ; Project Zero : discoverer ; an SSL certificate collision : vulnerability ; Kaspersky : vulnerable system owner ; Google : discoverer )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nBy Chris Duckett Google 's Project Zero has found that it was previously trivial to create an SSL certificate collision thanks to Kaspersky using only the first 32 bits of an MD5 hash in its SSL proxy packaged with its Anti-Virus product .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "discover vulnerability", "trigger": "has found", "arguments": [{"name": "Anti-Virus product", "role": "vulnerable system"}, {"name": "SSL proxy packaged", "role": "vulnerable system"}, {"name": "Project Zero", "role": "discoverer"}, {"name": "an SSL certificate collision", "role": "vulnerability"}, {"name": "Kaspersky", "role": "vulnerable system owner"}, {"name": "Google", "role": "discoverer"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( discover vulnerability : reported ; the bug : vulnerability ; Ormandy : discoverer )",
"id": "3713",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( discover vulnerability : reported ; the bug : vulnerability ; Ormandy : discoverer )",
"output_json": "{\"events\": [{\"event_type\": \"discover vulnerability\", \"trigger\": \"reported\", \"arguments\": [{\"name\": \"the bug\", \"role\": \"vulnerability\"}, {\"name\": \"Ormandy\", \"role\": \"discoverer\"}]}]}",
"sentence": "After Ormandy reported the bug and received acknowledgement from Kaspersky on November 1 , despite learning the security vendor was doing some commonName checks , the bug was still able to be exploited ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( discover vulnerability : reported ; the bug : vulnerability ; Ormandy : discoverer )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "discover vulnerability", "trigger": "reported", "arguments": [{"name": "the bug", "role": "vulnerability"}, {"name": "Ormandy", "role": "discoverer"}]}]}
|
( discover vulnerability : reported ; the bug : vulnerability ; Ormandy : discoverer )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nAfter Ormandy reported the bug and received acknowledgement from Kaspersky on November 1 , despite learning the security vendor was doing some commonName checks , the bug was still able to be exploited .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "discover vulnerability", "trigger": "reported", "arguments": [{"name": "the bug", "role": "vulnerability"}, {"name": "Ormandy", "role": "discoverer"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( discover vulnerability : also found ; another bug : vulnerability ; Ormandy : discoverer ; November 12 : time ; allowed any unprivileged user to become a local certificate authority : capabilities )",
"id": "3714",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( discover vulnerability : also found ; another bug : vulnerability ; Ormandy : discoverer ; November 12 : time ; allowed any unprivileged user to become a local certificate authority : capabilities )",
"output_json": "{\"events\": [{\"event_type\": \"discover vulnerability\", \"trigger\": \"also found\", \"arguments\": [{\"name\": \"another bug\", \"role\": \"vulnerability\"}, {\"name\": \"Ormandy\", \"role\": \"discoverer\"}, {\"name\": \"November 12\", \"role\": \"time\"}, {\"name\": \"allowed any unprivileged user to become a local certificate authority\", \"role\": \"capabilities\"}]}]}",
"sentence": "Ormandy also found another bug on November 12 that allowed any unprivileged user to become a local certificate authority ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( discover vulnerability : also found ; another bug : vulnerability ; Ormandy : discoverer ; November 12 : time ; allowed any unprivileged user to become a local certificate authority : capabilities )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "discover vulnerability", "trigger": "also found", "arguments": [{"name": "another bug", "role": "vulnerability"}, {"name": "Ormandy", "role": "discoverer"}, {"name": "November 12", "role": "time"}, {"name": "allowed any unprivileged user to become a local certificate authority", "role": "capabilities"}]}]}
|
( discover vulnerability : also found ; another bug : vulnerability ; Ormandy : discoverer ; November 12 : time ; allowed any unprivileged user to become a local certificate authority : capabilities )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nOrmandy also found another bug on November 12 that allowed any unprivileged user to become a local certificate authority .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "discover vulnerability", "trigger": "also found", "arguments": [{"name": "another bug", "role": "vulnerability"}, {"name": "Ormandy", "role": "discoverer"}, {"name": "November 12", "role": "time"}, {"name": "allowed any unprivileged user to become a local certificate authority", "role": "capabilities"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( discover vulnerability : discovered ; the Project Zero security researcher : discoverer ; Symantec Antivirus Engine : vulnerable system ; buffer overflow : vulnerability ; instant blue - screening : capabilities ; kernel memory corruption : capabilities ; May last year : time ; Windows : supported platform )",
"id": "3715",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( discover vulnerability : discovered ; the Project Zero security researcher : discoverer ; Symantec Antivirus Engine : vulnerable system ; buffer overflow : vulnerability ; instant blue - screening : capabilities ; kernel memory corruption : capabilities ; May last year : time ; Windows : supported platform )",
"output_json": "{\"events\": [{\"event_type\": \"discover vulnerability\", \"trigger\": \"discovered\", \"arguments\": [{\"name\": \"the Project Zero security researcher\", \"role\": \"discoverer\"}, {\"name\": \"Symantec Antivirus Engine\", \"role\": \"vulnerable system\"}, {\"name\": \"buffer overflow\", \"role\": \"vulnerability\"}, {\"name\": \"instant blue - screening\", \"role\": \"capabilities\"}, {\"name\": \"kernel memory corruption\", \"role\": \"capabilities\"}, {\"name\": \"May last year\", \"role\": \"time\"}, {\"name\": \"Windows\", \"role\": \"supported platform\"}]}]}",
"sentence": "In May last year , the Project Zero security researcher discovered that Symantec Antivirus Engine was vulnerable to buffer overflow when parsing malformed portable - executable header files that resulted in instant blue - screening and kernel memory corruption without user action on Windows ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( discover vulnerability : discovered ; the Project Zero security researcher : discoverer ; Symantec Antivirus Engine : vulnerable system ; buffer overflow : vulnerability ; instant blue - screening : capabilities ; kernel memory corruption : capabilities ; May last year : time ; Windows : supported platform )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "discover vulnerability", "trigger": "discovered", "arguments": [{"name": "the Project Zero security researcher", "role": "discoverer"}, {"name": "Symantec Antivirus Engine", "role": "vulnerable system"}, {"name": "buffer overflow", "role": "vulnerability"}, {"name": "instant blue - screening", "role": "capabilities"}, {"name": "kernel memory corruption", "role": "capabilities"}, {"name": "May last year", "role": "time"}, {"name": "Windows", "role": "supported platform"}]}]}
|
( discover vulnerability : discovered ; the Project Zero security researcher : discoverer ; Symantec Antivirus Engine : vulnerable system ; buffer overflow : vulnerability ; instant blue - screening : capabilities ; kernel memory corruption : capabilities ; May last year : time ; Windows : supported platform )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nIn May last year , the Project Zero security researcher discovered that Symantec Antivirus Engine was vulnerable to buffer overflow when parsing malformed portable - executable header files that resulted in instant blue - screening and kernel memory corruption without user action on Windows .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "discover vulnerability", "trigger": "discovered", "arguments": [{"name": "the Project Zero security researcher", "role": "discoverer"}, {"name": "Symantec Antivirus Engine", "role": "vulnerable system"}, {"name": "buffer overflow", "role": "vulnerability"}, {"name": "instant blue - screening", "role": "capabilities"}, {"name": "kernel memory corruption", "role": "capabilities"}, {"name": "May last year", "role": "time"}, {"name": "Windows", "role": "supported platform"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : collect ; location data : compromised data ; Ukrainian artillery : victim ; GRU : attacker ; custom - made malware : tool )",
"id": "3716",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : collect ; location data : compromised data ; Ukrainian artillery : victim ; GRU : attacker ; custom - made malware : tool )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"collect\", \"arguments\": [{\"name\": \"location data\", \"role\": \"compromised data\"}, {\"name\": \"Ukrainian artillery\", \"role\": \"victim\"}, {\"name\": \"GRU\", \"role\": \"attacker\"}, {\"name\": \"custom - made malware\", \"role\": \"tool\"}]}]}",
"sentence": "That ’s a surprising design flaw for custom - made malware whose alleged objective was to collect and transmit location data on Ukrainian artillery to the GRU ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : collect ; location data : compromised data ; Ukrainian artillery : victim ; GRU : attacker ; custom - made malware : tool )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "collect", "arguments": [{"name": "location data", "role": "compromised data"}, {"name": "Ukrainian artillery", "role": "victim"}, {"name": "GRU", "role": "attacker"}, {"name": "custom - made malware", "role": "tool"}]}]}
|
( data breach : collect ; location data : compromised data ; Ukrainian artillery : victim ; GRU : attacker ; custom - made malware : tool )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nThat ’s a surprising design flaw for custom - made malware whose alleged objective was to collect and transmit location data on Ukrainian artillery to the GRU .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "collect", "arguments": [{"name": "location data", "role": "compromised data"}, {"name": "Ukrainian artillery", "role": "victim"}, {"name": "GRU", "role": "attacker"}, {"name": "custom - made malware", "role": "tool"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : does collect ; base station information : compromised data ; It : attacker )",
"id": "3717",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : does collect ; base station information : compromised data ; It : attacker )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"does collect\", \"arguments\": [{\"name\": \"base station information\", \"role\": \"compromised data\"}, {\"name\": \"It\", \"role\": \"attacker\"}]}]}",
"sentence": "It does collect base station information but that is n’t nearly sufficient for targeting purposes ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : does collect ; base station information : compromised data ; It : attacker )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "does collect", "arguments": [{"name": "base station information", "role": "compromised data"}, {"name": "It", "role": "attacker"}]}]}
|
( data breach : does collect ; base station information : compromised data ; It : attacker )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nIt does collect base station information but that is n’t nearly sufficient for targeting purposes .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "does collect", "arguments": [{"name": "base station information", "role": "compromised data"}, {"name": "It", "role": "attacker"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : has been obtained ; the source code : compromised data ; others outside of Russia : attacker )",
"id": "3718",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : has been obtained ; the source code : compromised data ; others outside of Russia : attacker )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"has been obtained\", \"arguments\": [{\"name\": \"the source code\", \"role\": \"compromised data\"}, {\"name\": \"others outside of Russia\", \"role\": \"attacker\"}]}]}",
"sentence": "We now know that ’s false , and that the source code has been obtained by others outside of Russia ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : has been obtained ; the source code : compromised data ; others outside of Russia : attacker )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "has been obtained", "arguments": [{"name": "the source code", "role": "compromised data"}, {"name": "others outside of Russia", "role": "attacker"}]}]}
|
( data breach : has been obtained ; the source code : compromised data ; others outside of Russia : attacker )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nWe now know that ’s false , and that the source code has been obtained by others outside of Russia .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "has been obtained", "arguments": [{"name": "the source code", "role": "compromised data"}, {"name": "others outside of Russia", "role": "attacker"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( discover vulnerability : found ; artillery officer Yaroslav Sherstuk : vulnerable system owner ; researchers : discoverer ; an app : vulnerable system ; X-Agent : vulnerable system ; Android : supported platform )",
"id": "3719",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( discover vulnerability : found ; artillery officer Yaroslav Sherstuk : vulnerable system owner ; researchers : discoverer ; an app : vulnerable system ; X-Agent : vulnerable system ; Android : supported platform )",
"output_json": "{\"events\": [{\"event_type\": \"discover vulnerability\", \"trigger\": \"found\", \"arguments\": [{\"name\": \"artillery officer Yaroslav Sherstuk\", \"role\": \"vulnerable system owner\"}, {\"name\": \"researchers\", \"role\": \"discoverer\"}, {\"name\": \"an app\", \"role\": \"vulnerable system\"}, {\"name\": \"X-Agent\", \"role\": \"vulnerable system\"}, {\"name\": \"Android\", \"role\": \"supported platform\"}]}]}",
"sentence": "Specifically , researchers found an Android version of X-Agent , a piece of malware known to be used by Fancy Bear , embedded in an app developed by artillery officer Yaroslav Sherstuk to help military personnel reduce the time to fire D - 30 howitzers ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( discover vulnerability : found ; artillery officer Yaroslav Sherstuk : vulnerable system owner ; researchers : discoverer ; an app : vulnerable system ; X-Agent : vulnerable system ; Android : supported platform )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "discover vulnerability", "trigger": "found", "arguments": [{"name": "artillery officer Yaroslav Sherstuk", "role": "vulnerable system owner"}, {"name": "researchers", "role": "discoverer"}, {"name": "an app", "role": "vulnerable system"}, {"name": "X-Agent", "role": "vulnerable system"}, {"name": "Android", "role": "supported platform"}]}]}
|
( discover vulnerability : found ; artillery officer Yaroslav Sherstuk : vulnerable system owner ; researchers : discoverer ; an app : vulnerable system ; X-Agent : vulnerable system ; Android : supported platform )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nSpecifically , researchers found an Android version of X-Agent , a piece of malware known to be used by Fancy Bear , embedded in an app developed by artillery officer Yaroslav Sherstuk to help military personnel reduce the time to fire D - 30 howitzers .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "discover vulnerability", "trigger": "found", "arguments": [{"name": "artillery officer Yaroslav Sherstuk", "role": "vulnerable system owner"}, {"name": "researchers", "role": "discoverer"}, {"name": "an app", "role": "vulnerable system"}, {"name": "X-Agent", "role": "vulnerable system"}, {"name": "Android", "role": "supported platform"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : accessing ; the malicious app : tool ; forums : victim ; Ukrainian military : attacker ; late 2014 through 2016 : time ; contact information : compromised data ; SMS messages : compromised data ; call logs : compromised data ; Internet data : compromised data )",
"id": "3720",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : accessing ; the malicious app : tool ; forums : victim ; Ukrainian military : attacker ; late 2014 through 2016 : time ; contact information : compromised data ; SMS messages : compromised data ; call logs : compromised data ; Internet data : compromised data )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"accessing\", \"arguments\": [{\"name\": \"the malicious app\", \"role\": \"tool\"}, {\"name\": \"forums\", \"role\": \"victim\"}, {\"name\": \"Ukrainian military\", \"role\": \"attacker\"}, {\"name\": \"late 2014 through 2016\", \"role\": \"time\"}, {\"name\": \"contact information\", \"role\": \"compromised data\"}, {\"name\": \"SMS messages\", \"role\": \"compromised data\"}, {\"name\": \"call logs\", \"role\": \"compromised data\"}, {\"name\": \"Internet data\", \"role\": \"compromised data\"}]}]}",
"sentence": "According to CrowdStrike , the malicious app , which had been distributed on Ukrainian military forums from late 2014 through 2016 , was capable of accessing contact information , SMS messages , call logs and Internet data ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : accessing ; the malicious app : tool ; forums : victim ; Ukrainian military : attacker ; late 2014 through 2016 : time ; contact information : compromised data ; SMS messages : compromised data ; call logs : compromised data ; Internet data : compromised data )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "accessing", "arguments": [{"name": "the malicious app", "role": "tool"}, {"name": "forums", "role": "victim"}, {"name": "Ukrainian military", "role": "attacker"}, {"name": "late 2014 through 2016", "role": "time"}, {"name": "contact information", "role": "compromised data"}, {"name": "SMS messages", "role": "compromised data"}, {"name": "call logs", "role": "compromised data"}, {"name": "Internet data", "role": "compromised data"}]}]}
|
( data breach : accessing ; the malicious app : tool ; forums : victim ; Ukrainian military : attacker ; late 2014 through 2016 : time ; contact information : compromised data ; SMS messages : compromised data ; call logs : compromised data ; Internet data : compromised data )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nAccording to CrowdStrike , the malicious app , which had been distributed on Ukrainian military forums from late 2014 through 2016 , was capable of accessing contact information , SMS messages , call logs and Internet data .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "accessing", "arguments": [{"name": "the malicious app", "role": "tool"}, {"name": "forums", "role": "victim"}, {"name": "Ukrainian military", "role": "attacker"}, {"name": "late 2014 through 2016", "role": "time"}, {"name": "contact information", "role": "compromised data"}, {"name": "SMS messages", "role": "compromised data"}, {"name": "call logs", "role": "compromised data"}, {"name": "Internet data", "role": "compromised data"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : collect ; location data : compromised data ; Ukrainian artillery : victim ; custom - made malware : tool ; GRU : attacker )",
"id": "3721",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : collect ; location data : compromised data ; Ukrainian artillery : victim ; custom - made malware : tool ; GRU : attacker )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"collect\", \"arguments\": [{\"name\": \"location data\", \"role\": \"compromised data\"}, {\"name\": \"Ukrainian artillery\", \"role\": \"victim\"}, {\"name\": \"custom - made malware\", \"role\": \"tool\"}, {\"name\": \"GRU\", \"role\": \"attacker\"}]}]}",
"sentence": "Another interesting discovery is that the rogue app does not use GPS to obtain the infected device ’s exact location , which Carr names “ a surprising design flaw for custom - made malware whose alleged objective was to collect and transmit location data on Ukrainian artillery to the GRU ” ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : collect ; location data : compromised data ; Ukrainian artillery : victim ; custom - made malware : tool ; GRU : attacker )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "collect", "arguments": [{"name": "location data", "role": "compromised data"}, {"name": "Ukrainian artillery", "role": "victim"}, {"name": "custom - made malware", "role": "tool"}, {"name": "GRU", "role": "attacker"}]}]}
|
( data breach : collect ; location data : compromised data ; Ukrainian artillery : victim ; custom - made malware : tool ; GRU : attacker )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nAnother interesting discovery is that the rogue app does not use GPS to obtain the infected device ’s exact location , which Carr names “ a surprising design flaw for custom - made malware whose alleged objective was to collect and transmit location data on Ukrainian artillery to the GRU ” .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "collect", "arguments": [{"name": "location data", "role": "compromised data"}, {"name": "Ukrainian artillery", "role": "victim"}, {"name": "custom - made malware", "role": "tool"}, {"name": "GRU", "role": "attacker"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : collect ; location data : compromised data ; malware : tool ; via the base stations : attack pattern ; Android device : victim )",
"id": "3722",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : collect ; location data : compromised data ; malware : tool ; via the base stations : attack pattern ; Android device : victim )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"collect\", \"arguments\": [{\"name\": \"location data\", \"role\": \"compromised data\"}, {\"name\": \"malware\", \"role\": \"tool\"}, {\"name\": \"via the base stations\", \"role\": \"attack pattern\"}, {\"name\": \"Android device\", \"role\": \"victim\"}]}]}",
"sentence": "While the malware can collect some location data via the base stations used by the infected Android device , Carr believes it ’s not enough to track someone , especially given Ukraine ’s poor cellular service ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : collect ; location data : compromised data ; malware : tool ; via the base stations : attack pattern ; Android device : victim )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "collect", "arguments": [{"name": "location data", "role": "compromised data"}, {"name": "malware", "role": "tool"}, {"name": "via the base stations", "role": "attack pattern"}, {"name": "Android device", "role": "victim"}]}]}
|
( data breach : collect ; location data : compromised data ; malware : tool ; via the base stations : attack pattern ; Android device : victim )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nWhile the malware can collect some location data via the base stations used by the infected Android device , Carr believes it ’s not enough to track someone , especially given Ukraine ’s poor cellular service .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "collect", "arguments": [{"name": "location data", "role": "compromised data"}, {"name": "malware", "role": "tool"}, {"name": "via the base stations", "role": "attack pattern"}, {"name": "Android device", "role": "victim"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : a hacking campaign ; the group : attacker ; Democratic Party members : victim ; private emails : compromised data ; resulted in the publication : purpose ; destabilize the Hillary Clinton presidential campaign : purpose ; undermine the US democratic process : purpose )",
"id": "3723",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : a hacking campaign ; the group : attacker ; Democratic Party members : victim ; private emails : compromised data ; resulted in the publication : purpose ; destabilize the Hillary Clinton presidential campaign : purpose ; undermine the US democratic process : purpose )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"a hacking campaign\", \"arguments\": [{\"name\": \"the group\", \"role\": \"attacker\"}, {\"name\": \"Democratic Party members\", \"role\": \"victim\"}, {\"name\": \"private emails\", \"role\": \"compromised data\"}, {\"name\": \"resulted in the publication\", \"role\": \"purpose\"}, {\"name\": \"destabilize the Hillary Clinton presidential campaign\", \"role\": \"purpose\"}, {\"name\": \"undermine the US democratic process\", \"role\": \"purpose\"}]}]}",
"sentence": "More infamously , the group has been linked to a hacking campaign against Democratic Party members which resulted in the publication of private emails designed to destabilize the Hillary Clinton presidential campaign and undermine the US democratic process ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : a hacking campaign ; the group : attacker ; Democratic Party members : victim ; private emails : compromised data ; resulted in the publication : purpose ; destabilize the Hillary Clinton presidential campaign : purpose ; undermine the US democratic process : purpose )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "a hacking campaign", "arguments": [{"name": "the group", "role": "attacker"}, {"name": "Democratic Party members", "role": "victim"}, {"name": "private emails", "role": "compromised data"}, {"name": "resulted in the publication", "role": "purpose"}, {"name": "destabilize the Hillary Clinton presidential campaign", "role": "purpose"}, {"name": "undermine the US democratic process", "role": "purpose"}]}]}
|
( data breach : a hacking campaign ; the group : attacker ; Democratic Party members : victim ; private emails : compromised data ; resulted in the publication : purpose ; destabilize the Hillary Clinton presidential campaign : purpose ; undermine the US democratic process : purpose )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nMore infamously , the group has been linked to a hacking campaign against Democratic Party members which resulted in the publication of private emails designed to destabilize the Hillary Clinton presidential campaign and undermine the US democratic process .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "a hacking campaign", "arguments": [{"name": "the group", "role": "attacker"}, {"name": "Democratic Party members", "role": "victim"}, {"name": "private emails", "role": "compromised data"}, {"name": "resulted in the publication", "role": "purpose"}, {"name": "destabilize the Hillary Clinton presidential campaign", "role": "purpose"}, {"name": "undermine the US democratic process", "role": "purpose"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : leaked ; highly sensitive data : compromised data )",
"id": "3724",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : leaked ; highly sensitive data : compromised data )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"leaked\", \"arguments\": [{\"name\": \"highly sensitive data\", \"role\": \"compromised data\"}]}]}",
"sentence": "The hacker then hacked the Catalan Police Union server and leaked highly sensitive data against police brutality ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : leaked ; highly sensitive data : compromised data )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "leaked", "arguments": [{"name": "highly sensitive data", "role": "compromised data"}]}]}
|
( data breach : leaked ; highly sensitive data : compromised data )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nThe hacker then hacked the Catalan Police Union server and leaked highly sensitive data against police brutality .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "leaked", "arguments": [{"name": "highly sensitive data", "role": "compromised data"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : leaking ; personal details : compromised data ; police officers : victim )",
"id": "3725",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : leaking ; personal details : compromised data ; police officers : victim )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"leaking\", \"arguments\": [{\"name\": \"personal details\", \"role\": \"compromised data\"}, {\"name\": \"police officers\", \"role\": \"victim\"}]}]}",
"sentence": "The same hacker then hacked Spanish Police server and ended up leaking personal details of police officers against the infamous Gag Law ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : leaking ; personal details : compromised data ; police officers : victim )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "leaking", "arguments": [{"name": "personal details", "role": "compromised data"}, {"name": "police officers", "role": "victim"}]}]}
|
( data breach : leaking ; personal details : compromised data ; police officers : victim )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nThe same hacker then hacked Spanish Police server and ended up leaking personal details of police officers against the infamous Gag Law .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "leaking", "arguments": [{"name": "personal details", "role": "compromised data"}, {"name": "police officers", "role": "victim"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( discover vulnerability : explained ; the site : discoverer ; 26th December : time )",
"id": "3726",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( discover vulnerability : explained ; the site : discoverer ; 26th December : time )",
"output_json": "{\"events\": [{\"event_type\": \"discover vulnerability\", \"trigger\": \"explained\", \"arguments\": [{\"name\": \"the site\", \"role\": \"discoverer\"}, {\"name\": \"26th December\", \"role\": \"time\"}]}]}",
"sentence": "In an email sent to users on 26th December , the site explained that hackers were able to conduct this breach by exploiting a known vulnerability in outdated vBulletin forum software ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( discover vulnerability : explained ; the site : discoverer ; 26th December : time )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "discover vulnerability", "trigger": "explained", "arguments": [{"name": "the site", "role": "discoverer"}, {"name": "26th December", "role": "time"}]}]}
|
( discover vulnerability : explained ; the site : discoverer ; 26th December : time )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nIn an email sent to users on 26th December , the site explained that hackers were able to conduct this breach by exploiting a known vulnerability in outdated vBulletin forum software .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "discover vulnerability", "trigger": "explained", "arguments": [{"name": "the site", "role": "discoverer"}, {"name": "26th December", "role": "time"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : this breach ; users : victim ; users : victim )",
"id": "3727",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : this breach ; users : victim ; users : victim )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"this breach\", \"arguments\": [{\"name\": \"users\", \"role\": \"victim\"}, {\"name\": \"users\", \"role\": \"victim\"}]}]}",
"sentence": "Although PakWheels did n’t reveal the number of affected users , we at HackRead have inside details on this breach according to which the number of targeted users impacted by this breach goes over 674,775 users including names , emails , encrypted passwords , mobile number and Facebook sessions ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : this breach ; users : victim ; users : victim )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "this breach", "arguments": [{"name": "users", "role": "victim"}, {"name": "users", "role": "victim"}]}]}
|
( data breach : this breach ; users : victim ; users : victim )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nAlthough PakWheels did n’t reveal the number of affected users , we at HackRead have inside details on this breach according to which the number of targeted users impacted by this breach goes over 674,775 users including names , emails , encrypted passwords , mobile number and Facebook sessions .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "this breach", "arguments": [{"name": "users", "role": "victim"}, {"name": "users", "role": "victim"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : leaked ; database : compromised data ; Zameen : victim ; administration : victim )",
"id": "3728",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : leaked ; database : compromised data ; Zameen : victim ; administration : victim )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"leaked\", \"arguments\": [{\"name\": \"database\", \"role\": \"compromised data\"}, {\"name\": \"Zameen\", \"role\": \"victim\"}, {\"name\": \"administration\", \"role\": \"victim\"}]}]}",
"sentence": "In May 2016 , Pakistan ’s real estate giant Zameen was hacked by a Bangladeshi hacker who leaked its entire database after being ignored by Zameen ’s administration ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : leaked ; database : compromised data ; Zameen : victim ; administration : victim )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "leaked", "arguments": [{"name": "database", "role": "compromised data"}, {"name": "Zameen", "role": "victim"}, {"name": "administration", "role": "victim"}]}]}
|
( data breach : leaked ; database : compromised data ; Zameen : victim ; administration : victim )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nIn May 2016 , Pakistan ’s real estate giant Zameen was hacked by a Bangladeshi hacker who leaked its entire database after being ignored by Zameen ’s administration .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "leaked", "arguments": [{"name": "database", "role": "compromised data"}, {"name": "Zameen", "role": "victim"}, {"name": "administration", "role": "victim"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : the forums hacked ; data stolen : compromised data ; 1.6 million : number of victim ; Clash of Kings forum : victim ; outdated vBulletin forum software : victim ; accounts stolen : compromised data ; 12,000 : number of data ; Exile Mod gaming forum : victim ; accounts stolen : compromised data ; 7 million : number of data ; LifeBoat forum : victim ; 27 million : number of data ; Russia ’s Mail.ru : victim ; Grand Theft Auto ( GTA ) forum : victim ; accounts stolen : compromised data ; 800,000 : number of data ; Epic games forum : victim ; accounts stolen : compromised data )",
"id": "3729",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : the forums hacked ; data stolen : compromised data ; 1.6 million : number of victim ; Clash of Kings forum : victim ; outdated vBulletin forum software : victim ; accounts stolen : compromised data ; 12,000 : number of data ; Exile Mod gaming forum : victim ; accounts stolen : compromised data ; 7 million : number of data ; LifeBoat forum : victim ; 27 million : number of data ; Russia ’s Mail.ru : victim ; Grand Theft Auto ( GTA ) forum : victim ; accounts stolen : compromised data ; 800,000 : number of data ; Epic games forum : victim ; accounts stolen : compromised data )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"the forums hacked\", \"arguments\": [{\"name\": \"data stolen\", \"role\": \"compromised data\"}, {\"name\": \"1.6 million\", \"role\": \"number of victim\"}, {\"name\": \"Clash of Kings forum\", \"role\": \"victim\"}, {\"name\": \"outdated vBulletin forum software\", \"role\": \"victim\"}, {\"name\": \"accounts stolen\", \"role\": \"compromised data\"}, {\"name\": \"12,000\", \"role\": \"number of data\"}, {\"name\": \"Exile Mod gaming forum\", \"role\": \"victim\"}, {\"name\": \"accounts stolen\", \"role\": \"compromised data\"}, {\"name\": \"7 million\", \"role\": \"number of data\"}, {\"name\": \"LifeBoat forum\", \"role\": \"victim\"}, {\"name\": \"27 million\", \"role\": \"number of data\"}, {\"name\": \"Russia \\u2019s Mail.ru\", \"role\": \"victim\"}, {\"name\": \"Grand Theft Auto ( GTA ) forum\", \"role\": \"victim\"}, {\"name\": \"accounts stolen\", \"role\": \"compromised data\"}, {\"name\": \"800,000\", \"role\": \"number of data\"}, {\"name\": \"Epic games forum\", \"role\": \"victim\"}, {\"name\": \"accounts stolen\", \"role\": \"compromised data\"}]}]}",
"sentence": "Until now , the forums hacked due to vulnerability in outdated vBulletin forum software include Clash of Kings forum with 1.6 million data stolen , Epic games forum with 800,000 accounts stolen , Grand Theft Auto ( GTA ) forum , Russia ’s Mail.ru with 27 million accounts stolen , LifeBoat forum with 7 million accounts stolen and Exile Mod gaming forum with 12,000 accounts stolen ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : the forums hacked ; data stolen : compromised data ; 1.6 million : number of victim ; Clash of Kings forum : victim ; outdated vBulletin forum software : victim ; accounts stolen : compromised data ; 12,000 : number of data ; Exile Mod gaming forum : victim ; accounts stolen : compromised data ; 7 million : number of data ; LifeBoat forum : victim ; 27 million : number of data ; Russia ’s Mail.ru : victim ; Grand Theft Auto ( GTA ) forum : victim ; accounts stolen : compromised data ; 800,000 : number of data ; Epic games forum : victim ; accounts stolen : compromised data )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "the forums hacked", "arguments": [{"name": "data stolen", "role": "compromised data"}, {"name": "1.6 million", "role": "number of victim"}, {"name": "Clash of Kings forum", "role": "victim"}, {"name": "outdated vBulletin forum software", "role": "victim"}, {"name": "accounts stolen", "role": "compromised data"}, {"name": "12,000", "role": "number of data"}, {"name": "Exile Mod gaming forum", "role": "victim"}, {"name": "accounts stolen", "role": "compromised data"}, {"name": "7 million", "role": "number of data"}, {"name": "LifeBoat forum", "role": "victim"}, {"name": "27 million", "role": "number of data"}, {"name": "Russia \u2019s Mail.ru", "role": "victim"}, {"name": "Grand Theft Auto ( GTA ) forum", "role": "victim"}, {"name": "accounts stolen", "role": "compromised data"}, {"name": "800,000", "role": "number of data"}, {"name": "Epic games forum", "role": "victim"}, {"name": "accounts stolen", "role": "compromised data"}]}]}
|
( data breach : the forums hacked ; data stolen : compromised data ; 1.6 million : number of victim ; Clash of Kings forum : victim ; outdated vBulletin forum software : victim ; accounts stolen : compromised data ; 12,000 : number of data ; Exile Mod gaming forum : victim ; accounts stolen : compromised data ; 7 million : number of data ; LifeBoat forum : victim ; 27 million : number of data ; Russia ’s Mail.ru : victim ; Grand Theft Auto ( GTA ) forum : victim ; accounts stolen : compromised data ; 800,000 : number of data ; Epic games forum : victim ; accounts stolen : compromised data )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nUntil now , the forums hacked due to vulnerability in outdated vBulletin forum software include Clash of Kings forum with 1.6 million data stolen , Epic games forum with 800,000 accounts stolen , Grand Theft Auto ( GTA ) forum , Russia ’s Mail.ru with 27 million accounts stolen , LifeBoat forum with 7 million accounts stolen and Exile Mod gaming forum with 12,000 accounts stolen .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "the forums hacked", "arguments": [{"name": "data stolen", "role": "compromised data"}, {"name": "1.6 million", "role": "number of victim"}, {"name": "Clash of Kings forum", "role": "victim"}, {"name": "outdated vBulletin forum software", "role": "victim"}, {"name": "accounts stolen", "role": "compromised data"}, {"name": "12,000", "role": "number of data"}, {"name": "Exile Mod gaming forum", "role": "victim"}, {"name": "accounts stolen", "role": "compromised data"}, {"name": "7 million", "role": "number of data"}, {"name": "LifeBoat forum", "role": "victim"}, {"name": "27 million", "role": "number of data"}, {"name": "Russia \u2019s Mail.ru", "role": "victim"}, {"name": "Grand Theft Auto ( GTA ) forum", "role": "victim"}, {"name": "accounts stolen", "role": "compromised data"}, {"name": "800,000", "role": "number of data"}, {"name": "Epic games forum", "role": "victim"}, {"name": "accounts stolen", "role": "compromised data"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : published ; a database : compromised data ; the usernames : compromised data ; phone numbers : compromised data ; 4.6 million : number of victim ; Snapchat users : victim ; a group of hackers : attacker ; late 2013 : time )",
"id": "3730",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : published ; a database : compromised data ; the usernames : compromised data ; phone numbers : compromised data ; 4.6 million : number of victim ; Snapchat users : victim ; a group of hackers : attacker ; late 2013 : time )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"published\", \"arguments\": [{\"name\": \"a database\", \"role\": \"compromised data\"}, {\"name\": \"the usernames\", \"role\": \"compromised data\"}, {\"name\": \"phone numbers\", \"role\": \"compromised data\"}, {\"name\": \"4.6 million\", \"role\": \"number of victim\"}, {\"name\": \"Snapchat users\", \"role\": \"victim\"}, {\"name\": \"a group of hackers\", \"role\": \"attacker\"}, {\"name\": \"late 2013\", \"role\": \"time\"}]}]}",
"sentence": "For instance , back in late 2013 , a group of hackers published a database containing the usernames and phone numbers of approximately 4.6 million Snapchat users ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : published ; a database : compromised data ; the usernames : compromised data ; phone numbers : compromised data ; 4.6 million : number of victim ; Snapchat users : victim ; a group of hackers : attacker ; late 2013 : time )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "published", "arguments": [{"name": "a database", "role": "compromised data"}, {"name": "the usernames", "role": "compromised data"}, {"name": "phone numbers", "role": "compromised data"}, {"name": "4.6 million", "role": "number of victim"}, {"name": "Snapchat users", "role": "victim"}, {"name": "a group of hackers", "role": "attacker"}, {"name": "late 2013", "role": "time"}]}]}
|
( data breach : published ; a database : compromised data ; the usernames : compromised data ; phone numbers : compromised data ; 4.6 million : number of victim ; Snapchat users : victim ; a group of hackers : attacker ; late 2013 : time )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nFor instance , back in late 2013 , a group of hackers published a database containing the usernames and phone numbers of approximately 4.6 million Snapchat users .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "published", "arguments": [{"name": "a database", "role": "compromised data"}, {"name": "the usernames", "role": "compromised data"}, {"name": "phone numbers", "role": "compromised data"}, {"name": "4.6 million", "role": "number of victim"}, {"name": "Snapchat users", "role": "victim"}, {"name": "a group of hackers", "role": "attacker"}, {"name": "late 2013", "role": "time"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( phishing : the security incident ; February 2016 : time )",
"id": "3731",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( phishing : the security incident ; February 2016 : time )",
"output_json": "{\"events\": [{\"event_type\": \"phishing\", \"trigger\": \"the security incident\", \"arguments\": [{\"name\": \"February 2016\", \"role\": \"time\"}]}]}",
"sentence": "We also ca n't forget about the security incident that occurred back in February 2016 ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( phishing : the security incident ; February 2016 : time )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "phishing", "trigger": "the security incident", "arguments": [{"name": "February 2016", "role": "time"}]}]}
|
( phishing : the security incident ; February 2016 : time )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nWe also ca n't forget about the security incident that occurred back in February 2016 .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "phishing", "trigger": "the security incident", "arguments": [{"name": "February 2016", "role": "time"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( phishing : that attack ; None )",
"id": "3732",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( phishing : that attack ; None )",
"output_json": "{\"events\": [{\"event_type\": \"phishing\", \"trigger\": \"that attack\", \"arguments\": []}]}",
"sentence": "In that attack , someone posed as the company 's CEO and convinced a Snapchat employee to send over payroll information ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( phishing : that attack ; None )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "phishing", "trigger": "that attack", "arguments": []}]}
|
( phishing : that attack ; None )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nIn that attack , someone posed as the company 's CEO and convinced a Snapchat employee to send over payroll information .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "phishing", "trigger": "that attack", "arguments": []}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : compromised ; identities : compromised data ; dozens : number of victim ; The successful phish : attack pattern ; employees : victim )",
"id": "3733",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : compromised ; identities : compromised data ; dozens : number of victim ; The successful phish : attack pattern ; employees : victim )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"compromised\", \"arguments\": [{\"name\": \"identities\", \"role\": \"compromised data\"}, {\"name\": \"dozens\", \"role\": \"number of victim\"}, {\"name\": \"The successful phish\", \"role\": \"attack pattern\"}, {\"name\": \"employees\", \"role\": \"victim\"}]}]}",
"sentence": "The successful phish ultimately compromised dozens of employees ' identities ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : compromised ; identities : compromised data ; dozens : number of victim ; The successful phish : attack pattern ; employees : victim )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "compromised", "arguments": [{"name": "identities", "role": "compromised data"}, {"name": "dozens", "role": "number of victim"}, {"name": "The successful phish", "role": "attack pattern"}, {"name": "employees", "role": "victim"}]}]}
|
( data breach : compromised ; identities : compromised data ; dozens : number of victim ; The successful phish : attack pattern ; employees : victim )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nThe successful phish ultimately compromised dozens of employees ' identities .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "compromised", "arguments": [{"name": "identities", "role": "compromised data"}, {"name": "dozens", "role": "number of victim"}, {"name": "The successful phish", "role": "attack pattern"}, {"name": "employees", "role": "victim"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( phishing : claim to be ; they : victim ; Android users : victim ; the Super Mario Run : trusted entity ; the apps : trusted entity )",
"id": "3734",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( phishing : claim to be ; they : victim ; Android users : victim ; the Super Mario Run : trusted entity ; the apps : trusted entity )",
"output_json": "{\"events\": [{\"event_type\": \"phishing\", \"trigger\": \"claim to be\", \"arguments\": [{\"name\": \"they\", \"role\": \"victim\"}, {\"name\": \"Android users\", \"role\": \"victim\"}, {\"name\": \"the Super Mario Run\", \"role\": \"trusted entity\"}, {\"name\": \"the apps\", \"role\": \"trusted entity\"}]}]}",
"sentence": "Warnings on Fake Super Mario Android Apps Recent warnings have been given by security experts for Android users that they should be aware of the apps that falsely claim to be the Super Mario Run for their device ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( phishing : claim to be ; they : victim ; Android users : victim ; the Super Mario Run : trusted entity ; the apps : trusted entity )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "phishing", "trigger": "claim to be", "arguments": [{"name": "they", "role": "victim"}, {"name": "Android users", "role": "victim"}, {"name": "the Super Mario Run", "role": "trusted entity"}, {"name": "the apps", "role": "trusted entity"}]}]}
|
( phishing : claim to be ; they : victim ; Android users : victim ; the Super Mario Run : trusted entity ; the apps : trusted entity )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nWarnings on Fake Super Mario Android Apps Recent warnings have been given by security experts for Android users that they should be aware of the apps that falsely claim to be the Super Mario Run for their device .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "phishing", "trigger": "claim to be", "arguments": [{"name": "they", "role": "victim"}, {"name": "Android users", "role": "victim"}, {"name": "the Super Mario Run", "role": "trusted entity"}, {"name": "the apps", "role": "trusted entity"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( phishing : fake downloads ; uncovering an actual mountain of malware : purpose ; the game : trusted entity )",
"id": "3735",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( phishing : fake downloads ; uncovering an actual mountain of malware : purpose ; the game : trusted entity )",
"output_json": "{\"events\": [{\"event_type\": \"phishing\", \"trigger\": \"fake downloads\", \"arguments\": [{\"name\": \"uncovering an actual mountain of malware\", \"role\": \"purpose\"}, {\"name\": \"the game\", \"role\": \"trusted entity\"}]}]}",
"sentence": "However , such fake downloads end up uncovering an actual mountain of malware instead of the game ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( phishing : fake downloads ; uncovering an actual mountain of malware : purpose ; the game : trusted entity )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "phishing", "trigger": "fake downloads", "arguments": [{"name": "uncovering an actual mountain of malware", "role": "purpose"}, {"name": "the game", "role": "trusted entity"}]}]}
|
( phishing : fake downloads ; uncovering an actual mountain of malware : purpose ; the game : trusted entity )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nHowever , such fake downloads end up uncovering an actual mountain of malware instead of the game .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "phishing", "trigger": "fake downloads", "arguments": [{"name": "uncovering an actual mountain of malware", "role": "purpose"}, {"name": "the game", "role": "trusted entity"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( phishing : claim as ; other content : trusted entity ; game Super Mario Run : trusted entity ; fake Super Mario Run Android apps : trusted entity )",
"id": "3736",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( phishing : claim as ; other content : trusted entity ; game Super Mario Run : trusted entity ; fake Super Mario Run Android apps : trusted entity )",
"output_json": "{\"events\": [{\"event_type\": \"phishing\", \"trigger\": \"claim as\", \"arguments\": [{\"name\": \"other content\", \"role\": \"trusted entity\"}, {\"name\": \"game Super Mario Run\", \"role\": \"trusted entity\"}, {\"name\": \"fake Super Mario Run Android apps\", \"role\": \"trusted entity\"}]}]}",
"sentence": "Trend Micro Findings on Malicious Apps Security Firm Trend Micro 's Researchers have recently found an increasing count of fake Super Mario Run Android apps that claim as the trending game Super Mario Run or other content related to it ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( phishing : claim as ; other content : trusted entity ; game Super Mario Run : trusted entity ; fake Super Mario Run Android apps : trusted entity )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "phishing", "trigger": "claim as", "arguments": [{"name": "other content", "role": "trusted entity"}, {"name": "game Super Mario Run", "role": "trusted entity"}, {"name": "fake Super Mario Run Android apps", "role": "trusted entity"}]}]}
|
( phishing : claim as ; other content : trusted entity ; game Super Mario Run : trusted entity ; fake Super Mario Run Android apps : trusted entity )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nTrend Micro Findings on Malicious Apps Security Firm Trend Micro 's Researchers have recently found an increasing count of fake Super Mario Run Android apps that claim as the trending game Super Mario Run or other content related to it .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "phishing", "trigger": "claim as", "arguments": [{"name": "other content", "role": "trusted entity"}, {"name": "game Super Mario Run", "role": "trusted entity"}, {"name": "fake Super Mario Run Android apps", "role": "trusted entity"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( ransom : were infected ; computer systems : victim ; ransomware : tool ; pharmaceutical company : victim ; Indian banks : victim ; January 2016 : time )",
"id": "3737",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( ransom : were infected ; computer systems : victim ; ransomware : tool ; pharmaceutical company : victim ; Indian banks : victim ; January 2016 : time )",
"output_json": "{\"events\": [{\"event_type\": \"ransom\", \"trigger\": \"were infected\", \"arguments\": [{\"name\": \"computer systems\", \"role\": \"victim\"}, {\"name\": \"ransomware\", \"role\": \"tool\"}, {\"name\": \"pharmaceutical company\", \"role\": \"victim\"}, {\"name\": \"Indian banks\", \"role\": \"victim\"}, {\"name\": \"January 2016\", \"role\": \"time\"}]}]}",
"sentence": "The 2016 Ransomware Report released by 360 Security Center lately presents that : – 4.9 million computers were attacked in China – 56,000 ransomware infections worldwide only in March 2016 – $ 1 billion dollar source of income for cyber criminals estimated by FBI – Almost half of organizations have been hit with ransomware In January 2016 , three Indian banks ’ and a pharmaceutical company ’s computer systems were infected by ransomware ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( ransom : were infected ; computer systems : victim ; ransomware : tool ; pharmaceutical company : victim ; Indian banks : victim ; January 2016 : time )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "ransom", "trigger": "were infected", "arguments": [{"name": "computer systems", "role": "victim"}, {"name": "ransomware", "role": "tool"}, {"name": "pharmaceutical company", "role": "victim"}, {"name": "Indian banks", "role": "victim"}, {"name": "January 2016", "role": "time"}]}]}
|
( ransom : were infected ; computer systems : victim ; ransomware : tool ; pharmaceutical company : victim ; Indian banks : victim ; January 2016 : time )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nThe 2016 Ransomware Report released by 360 Security Center lately presents that : – 4.9 million computers were attacked in China – 56,000 ransomware infections worldwide only in March 2016 – $ 1 billion dollar source of income for cyber criminals estimated by FBI – Almost half of organizations have been hit with ransomware In January 2016 , three Indian banks ’ and a pharmaceutical company ’s computer systems were infected by ransomware .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "ransom", "trigger": "were infected", "arguments": [{"name": "computer systems", "role": "victim"}, {"name": "ransomware", "role": "tool"}, {"name": "pharmaceutical company", "role": "victim"}, {"name": "Indian banks", "role": "victim"}, {"name": "January 2016", "role": "time"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( ransom : asked for ; 1 bitcoin : price ; attacker : attacker ; $ 905 : price ; infected computer : victim ; unprotected desktop interface : victim ; infect other connected computers from remote : attack pattern )",
"id": "3738",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( ransom : asked for ; 1 bitcoin : price ; attacker : attacker ; $ 905 : price ; infected computer : victim ; unprotected desktop interface : victim ; infect other connected computers from remote : attack pattern )",
"output_json": "{\"events\": [{\"event_type\": \"ransom\", \"trigger\": \"asked for\", \"arguments\": [{\"name\": \"1 bitcoin\", \"role\": \"price\"}, {\"name\": \"attacker\", \"role\": \"attacker\"}, {\"name\": \"$ 905\", \"role\": \"price\"}, {\"name\": \"infected computer\", \"role\": \"victim\"}, {\"name\": \"unprotected desktop interface\", \"role\": \"victim\"}, {\"name\": \"infect other connected computers from remote\", \"role\": \"attack pattern\"}]}]}",
"sentence": "The attacker asked for 1 bitcoin ( about $ 905 ) for each infected computer , and then used unprotected desktop interface to infect other connected computers from remote ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( ransom : asked for ; 1 bitcoin : price ; attacker : attacker ; $ 905 : price ; infected computer : victim ; unprotected desktop interface : victim ; infect other connected computers from remote : attack pattern )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "ransom", "trigger": "asked for", "arguments": [{"name": "1 bitcoin", "role": "price"}, {"name": "attacker", "role": "attacker"}, {"name": "$ 905", "role": "price"}, {"name": "infected computer", "role": "victim"}, {"name": "unprotected desktop interface", "role": "victim"}, {"name": "infect other connected computers from remote", "role": "attack pattern"}]}]}
|
( ransom : asked for ; 1 bitcoin : price ; attacker : attacker ; $ 905 : price ; infected computer : victim ; unprotected desktop interface : victim ; infect other connected computers from remote : attack pattern )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nThe attacker asked for 1 bitcoin ( about $ 905 ) for each infected computer , and then used unprotected desktop interface to infect other connected computers from remote .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "ransom", "trigger": "asked for", "arguments": [{"name": "1 bitcoin", "role": "price"}, {"name": "attacker", "role": "attacker"}, {"name": "$ 905", "role": "price"}, {"name": "infected computer", "role": "victim"}, {"name": "unprotected desktop interface", "role": "victim"}, {"name": "infect other connected computers from remote", "role": "attack pattern"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( ransom : paid ; February 5th 2016 : time ; Hollywood Presbyterian Medical Center : victim ; $ 17,000 : price ; in bitcoin : payment method ; hacker : attacker ; seized control : attack pattern ; computer systems : victim ; hospital : victim )",
"id": "3739",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( ransom : paid ; February 5th 2016 : time ; Hollywood Presbyterian Medical Center : victim ; $ 17,000 : price ; in bitcoin : payment method ; hacker : attacker ; seized control : attack pattern ; computer systems : victim ; hospital : victim )",
"output_json": "{\"events\": [{\"event_type\": \"ransom\", \"trigger\": \"paid\", \"arguments\": [{\"name\": \"February 5th 2016\", \"role\": \"time\"}, {\"name\": \"Hollywood Presbyterian Medical Center\", \"role\": \"victim\"}, {\"name\": \"$ 17,000\", \"role\": \"price\"}, {\"name\": \"in bitcoin\", \"role\": \"payment method\"}, {\"name\": \"hacker\", \"role\": \"attacker\"}, {\"name\": \"seized control\", \"role\": \"attack pattern\"}, {\"name\": \"computer systems\", \"role\": \"victim\"}, {\"name\": \"hospital\", \"role\": \"victim\"}]}]}",
"sentence": "February 5th 2016 , Hollywood Presbyterian Medical Center paid a $ 17,000 ransom in bitcoin to a hacker who seized control of the hospital ’s computer systems and would give back access only when the money was paid ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( ransom : paid ; February 5th 2016 : time ; Hollywood Presbyterian Medical Center : victim ; $ 17,000 : price ; in bitcoin : payment method ; hacker : attacker ; seized control : attack pattern ; computer systems : victim ; hospital : victim )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "ransom", "trigger": "paid", "arguments": [{"name": "February 5th 2016", "role": "time"}, {"name": "Hollywood Presbyterian Medical Center", "role": "victim"}, {"name": "$ 17,000", "role": "price"}, {"name": "in bitcoin", "role": "payment method"}, {"name": "hacker", "role": "attacker"}, {"name": "seized control", "role": "attack pattern"}, {"name": "computer systems", "role": "victim"}, {"name": "hospital", "role": "victim"}]}]}
|
( ransom : paid ; February 5th 2016 : time ; Hollywood Presbyterian Medical Center : victim ; $ 17,000 : price ; in bitcoin : payment method ; hacker : attacker ; seized control : attack pattern ; computer systems : victim ; hospital : victim )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nFebruary 5th 2016 , Hollywood Presbyterian Medical Center paid a $ 17,000 ransom in bitcoin to a hacker who seized control of the hospital ’s computer systems and would give back access only when the money was paid .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "ransom", "trigger": "paid", "arguments": [{"name": "February 5th 2016", "role": "time"}, {"name": "Hollywood Presbyterian Medical Center", "role": "victim"}, {"name": "$ 17,000", "role": "price"}, {"name": "in bitcoin", "role": "payment method"}, {"name": "hacker", "role": "attacker"}, {"name": "seized control", "role": "attack pattern"}, {"name": "computer systems", "role": "victim"}, {"name": "hospital", "role": "victim"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( ransom : asked for ; servers : victim ; intranet : victim ; hacker : attacker ; 20 bitcoin : price ; took control : attack pattern )",
"id": "3740",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( ransom : asked for ; servers : victim ; intranet : victim ; hacker : attacker ; 20 bitcoin : price ; took control : attack pattern )",
"output_json": "{\"events\": [{\"event_type\": \"ransom\", \"trigger\": \"asked for\", \"arguments\": [{\"name\": \"servers\", \"role\": \"victim\"}, {\"name\": \"intranet\", \"role\": \"victim\"}, {\"name\": \"hacker\", \"role\": \"attacker\"}, {\"name\": \"20 bitcoin\", \"role\": \"price\"}, {\"name\": \"took control\", \"role\": \"attack pattern\"}]}]}",
"sentence": "The hacker took control of the intranet and servers , and asked for 20 bitcoin ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( ransom : asked for ; servers : victim ; intranet : victim ; hacker : attacker ; 20 bitcoin : price ; took control : attack pattern )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "ransom", "trigger": "asked for", "arguments": [{"name": "servers", "role": "victim"}, {"name": "intranet", "role": "victim"}, {"name": "hacker", "role": "attacker"}, {"name": "20 bitcoin", "role": "price"}, {"name": "took control", "role": "attack pattern"}]}]}
|
( ransom : asked for ; servers : victim ; intranet : victim ; hacker : attacker ; 20 bitcoin : price ; took control : attack pattern )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nThe hacker took control of the intranet and servers , and asked for 20 bitcoin .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "ransom", "trigger": "asked for", "arguments": [{"name": "servers", "role": "victim"}, {"name": "intranet", "role": "victim"}, {"name": "hacker", "role": "attacker"}, {"name": "20 bitcoin", "role": "price"}, {"name": "took control", "role": "attack pattern"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( ransom : ended up paying ; hacker : attacker ; $ 8,500 : price ; school : victim ; IT systems : victim )",
"id": "3741",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( ransom : ended up paying ; hacker : attacker ; $ 8,500 : price ; school : victim ; IT systems : victim )",
"output_json": "{\"events\": [{\"event_type\": \"ransom\", \"trigger\": \"ended up paying\", \"arguments\": [{\"name\": \"hacker\", \"role\": \"attacker\"}, {\"name\": \"$ 8,500\", \"role\": \"price\"}, {\"name\": \"school\", \"role\": \"victim\"}, {\"name\": \"IT systems\", \"role\": \"victim\"}]}]}",
"sentence": "These school ended up paying the anonymous hacker $ 8,500 to get their IT systems back ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( ransom : ended up paying ; hacker : attacker ; $ 8,500 : price ; school : victim ; IT systems : victim )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "ransom", "trigger": "ended up paying", "arguments": [{"name": "hacker", "role": "attacker"}, {"name": "$ 8,500", "role": "price"}, {"name": "school", "role": "victim"}, {"name": "IT systems", "role": "victim"}]}]}
|
( ransom : ended up paying ; hacker : attacker ; $ 8,500 : price ; school : victim ; IT systems : victim )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nThese school ended up paying the anonymous hacker $ 8,500 to get their IT systems back .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "ransom", "trigger": "ended up paying", "arguments": [{"name": "hacker", "role": "attacker"}, {"name": "$ 8,500", "role": "price"}, {"name": "school", "role": "victim"}, {"name": "IT systems", "role": "victim"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( ransom : had to pay ; their files were encrypted : attack pattern ; the distributor : attacker ; users : victim )",
"id": "3742",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( ransom : had to pay ; their files were encrypted : attack pattern ; the distributor : attacker ; users : victim )",
"output_json": "{\"events\": [{\"event_type\": \"ransom\", \"trigger\": \"had to pay\", \"arguments\": [{\"name\": \"their files were encrypted\", \"role\": \"attack pattern\"}, {\"name\": \"the distributor\", \"role\": \"attacker\"}, {\"name\": \"users\", \"role\": \"victim\"}]}]}",
"sentence": "Once users activated the file attached in the email , their files were encrypted and had to pay the distributor a certain ransom to decrypt these files ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( ransom : had to pay ; their files were encrypted : attack pattern ; the distributor : attacker ; users : victim )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "ransom", "trigger": "had to pay", "arguments": [{"name": "their files were encrypted", "role": "attack pattern"}, {"name": "the distributor", "role": "attacker"}, {"name": "users", "role": "victim"}]}]}
|
( ransom : had to pay ; their files were encrypted : attack pattern ; the distributor : attacker ; users : victim )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nOnce users activated the file attached in the email , their files were encrypted and had to pay the distributor a certain ransom to decrypt these files .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "ransom", "trigger": "had to pay", "arguments": [{"name": "their files were encrypted", "role": "attack pattern"}, {"name": "the distributor", "role": "attacker"}, {"name": "users", "role": "victim"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( ransom : ransomware attacks ; October , 2016 : time ; Hong Kong : place ; China : place )",
"id": "3743",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( ransom : ransomware attacks ; October , 2016 : time ; Hong Kong : place ; China : place )",
"output_json": "{\"events\": [{\"event_type\": \"ransom\", \"trigger\": \"ransomware attacks\", \"arguments\": [{\"name\": \"October , 2016\", \"role\": \"time\"}, {\"name\": \"Hong Kong\", \"role\": \"place\"}, {\"name\": \"China\", \"role\": \"place\"}]}]}",
"sentence": "In October , 2016 , 277 ransomware attacks were reported to Government Computer Emergency Response Team in Hong Kong , China ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( ransom : ransomware attacks ; October , 2016 : time ; Hong Kong : place ; China : place )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "ransom", "trigger": "ransomware attacks", "arguments": [{"name": "October , 2016", "role": "time"}, {"name": "Hong Kong", "role": "place"}, {"name": "China", "role": "place"}]}]}
|
( ransom : ransomware attacks ; October , 2016 : time ; Hong Kong : place ; China : place )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nIn October , 2016 , 277 ransomware attacks were reported to Government Computer Emergency Response Team in Hong Kong , China .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "ransom", "trigger": "ransomware attacks", "arguments": [{"name": "October , 2016", "role": "time"}, {"name": "Hong Kong", "role": "place"}, {"name": "China", "role": "place"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( phishing : disguised as ; bills : trusted entity ; receipts : trusted entity ; email attachments : tool ; the malware : tool )",
"id": "3744",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( phishing : disguised as ; bills : trusted entity ; receipts : trusted entity ; email attachments : tool ; the malware : tool )",
"output_json": "{\"events\": [{\"event_type\": \"phishing\", \"trigger\": \"disguised as\", \"arguments\": [{\"name\": \"bills\", \"role\": \"trusted entity\"}, {\"name\": \"receipts\", \"role\": \"trusted entity\"}, {\"name\": \"email attachments\", \"role\": \"tool\"}, {\"name\": \"the malware\", \"role\": \"tool\"}]}]}",
"sentence": "Most of the malware were hidden in email attachments and disguised as bills or receipts to trick users to click ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( phishing : disguised as ; bills : trusted entity ; receipts : trusted entity ; email attachments : tool ; the malware : tool )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "phishing", "trigger": "disguised as", "arguments": [{"name": "bills", "role": "trusted entity"}, {"name": "receipts", "role": "trusted entity"}, {"name": "email attachments", "role": "tool"}, {"name": "the malware", "role": "tool"}]}]}
|
( phishing : disguised as ; bills : trusted entity ; receipts : trusted entity ; email attachments : tool ; the malware : tool )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nMost of the malware were hidden in email attachments and disguised as bills or receipts to trick users to click .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "phishing", "trigger": "disguised as", "arguments": [{"name": "bills", "role": "trusted entity"}, {"name": "receipts", "role": "trusted entity"}, {"name": "email attachments", "role": "tool"}, {"name": "the malware", "role": "tool"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( ransom : requested for ; $ 73,000 : price ; in bitcoin : payment method ; San Francisco public transportation system Muni : victim ; November 2016 : time )",
"id": "3745",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( ransom : requested for ; $ 73,000 : price ; in bitcoin : payment method ; San Francisco public transportation system Muni : victim ; November 2016 : time )",
"output_json": "{\"events\": [{\"event_type\": \"ransom\", \"trigger\": \"requested for\", \"arguments\": [{\"name\": \"$ 73,000\", \"role\": \"price\"}, {\"name\": \"in bitcoin\", \"role\": \"payment method\"}, {\"name\": \"San Francisco public transportation system Muni\", \"role\": \"victim\"}, {\"name\": \"November 2016\", \"role\": \"time\"}]}]}",
"sentence": "November 2016 , San Francisco public transportation system Muni was hacked and requested for a $ 73,000 ransom in bitcoin to get back encrypted data ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( ransom : requested for ; $ 73,000 : price ; in bitcoin : payment method ; San Francisco public transportation system Muni : victim ; November 2016 : time )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "ransom", "trigger": "requested for", "arguments": [{"name": "$ 73,000", "role": "price"}, {"name": "in bitcoin", "role": "payment method"}, {"name": "San Francisco public transportation system Muni", "role": "victim"}, {"name": "November 2016", "role": "time"}]}]}
|
( ransom : requested for ; $ 73,000 : price ; in bitcoin : payment method ; San Francisco public transportation system Muni : victim ; November 2016 : time )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nNovember 2016 , San Francisco public transportation system Muni was hacked and requested for a $ 73,000 ransom in bitcoin to get back encrypted data .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "ransom", "trigger": "requested for", "arguments": [{"name": "$ 73,000", "role": "price"}, {"name": "in bitcoin", "role": "payment method"}, {"name": "San Francisco public transportation system Muni", "role": "victim"}, {"name": "November 2016", "role": "time"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( ransom : refused to pay ; SFMTA ( The San Francisco Municipal Transportation Authority ) : victim ; the fair system : victim )",
"id": "3746",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( ransom : refused to pay ; SFMTA ( The San Francisco Municipal Transportation Authority ) : victim ; the fair system : victim )",
"output_json": "{\"events\": [{\"event_type\": \"ransom\", \"trigger\": \"refused to pay\", \"arguments\": [{\"name\": \"SFMTA ( The San Francisco Municipal Transportation Authority )\", \"role\": \"victim\"}, {\"name\": \"the fair system\", \"role\": \"victim\"}]}]}",
"sentence": "SFMTA ( The San Francisco Municipal Transportation Authority ) refused to pay the ransom and shut down the fair system ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( ransom : refused to pay ; SFMTA ( The San Francisco Municipal Transportation Authority ) : victim ; the fair system : victim )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "ransom", "trigger": "refused to pay", "arguments": [{"name": "SFMTA ( The San Francisco Municipal Transportation Authority )", "role": "victim"}, {"name": "the fair system", "role": "victim"}]}]}
|
( ransom : refused to pay ; SFMTA ( The San Francisco Municipal Transportation Authority ) : victim ; the fair system : victim )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nSFMTA ( The San Francisco Municipal Transportation Authority ) refused to pay the ransom and shut down the fair system .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "ransom", "trigger": "refused to pay", "arguments": [{"name": "SFMTA ( The San Francisco Municipal Transportation Authority )", "role": "victim"}, {"name": "the fair system", "role": "victim"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( discover vulnerability : said ; security vulnerabilities : vulnerability ; supervisory control and data acquisition systems : vulnerable system ; remotely control critical infrastructure : capabilities ; The Russian firm : discoverer )",
"id": "3747",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( discover vulnerability : said ; security vulnerabilities : vulnerability ; supervisory control and data acquisition systems : vulnerable system ; remotely control critical infrastructure : capabilities ; The Russian firm : discoverer )",
"output_json": "{\"events\": [{\"event_type\": \"discover vulnerability\", \"trigger\": \"said\", \"arguments\": [{\"name\": \"security vulnerabilities\", \"role\": \"vulnerability\"}, {\"name\": \"supervisory control and data acquisition systems\", \"role\": \"vulnerable system\"}, {\"name\": \"remotely control critical infrastructure\", \"role\": \"capabilities\"}, {\"name\": \"The Russian firm\", \"role\": \"discoverer\"}]}]}",
"sentence": "The Russian firm is said to have extensive technical experience in security vulnerabilities of supervisory control and data acquisition systems that are used to remotely control critical infrastructure ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( discover vulnerability : said ; security vulnerabilities : vulnerability ; supervisory control and data acquisition systems : vulnerable system ; remotely control critical infrastructure : capabilities ; The Russian firm : discoverer )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "discover vulnerability", "trigger": "said", "arguments": [{"name": "security vulnerabilities", "role": "vulnerability"}, {"name": "supervisory control and data acquisition systems", "role": "vulnerable system"}, {"name": "remotely control critical infrastructure", "role": "capabilities"}, {"name": "The Russian firm", "role": "discoverer"}]}]}
|
( discover vulnerability : said ; security vulnerabilities : vulnerability ; supervisory control and data acquisition systems : vulnerable system ; remotely control critical infrastructure : capabilities ; The Russian firm : discoverer )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nThe Russian firm is said to have extensive technical experience in security vulnerabilities of supervisory control and data acquisition systems that are used to remotely control critical infrastructure .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "discover vulnerability", "trigger": "said", "arguments": [{"name": "security vulnerabilities", "role": "vulnerability"}, {"name": "supervisory control and data acquisition systems", "role": "vulnerable system"}, {"name": "remotely control critical infrastructure", "role": "capabilities"}, {"name": "The Russian firm", "role": "discoverer"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : access ; DNC emails : compromised data ; the spear - fishing process : attack pattern ; Grizzly Steppe : attack pattern )",
"id": "3748",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : access ; DNC emails : compromised data ; the spear - fishing process : attack pattern ; Grizzly Steppe : attack pattern )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"access\", \"arguments\": [{\"name\": \"DNC emails\", \"role\": \"compromised data\"}, {\"name\": \"the spear - fishing process\", \"role\": \"attack pattern\"}, {\"name\": \"Grizzly Steppe\", \"role\": \"attack pattern\"}]}]}",
"sentence": "The recent political furor over state sponsored hacking took an ugly and dangerous turn , on the morning of December 30th when a tiny Vermont electric utility reported that Grizzly Steppe – the spear - fishing process used to access DNC emails – had been found on one of their systems ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : access ; DNC emails : compromised data ; the spear - fishing process : attack pattern ; Grizzly Steppe : attack pattern )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "access", "arguments": [{"name": "DNC emails", "role": "compromised data"}, {"name": "the spear - fishing process", "role": "attack pattern"}, {"name": "Grizzly Steppe", "role": "attack pattern"}]}]}
|
( data breach : access ; DNC emails : compromised data ; the spear - fishing process : attack pattern ; Grizzly Steppe : attack pattern )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nThe recent political furor over state sponsored hacking took an ugly and dangerous turn , on the morning of December 30th when a tiny Vermont electric utility reported that Grizzly Steppe – the spear - fishing process used to access DNC emails – had been found on one of their systems .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "access", "arguments": [{"name": "DNC emails", "role": "compromised data"}, {"name": "the spear - fishing process", "role": "attack pattern"}, {"name": "Grizzly Steppe", "role": "attack pattern"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( data breach : acquire ; login credentials : compromised data ; hackers : attacker ; Two years prior : time )",
"id": "3749",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( data breach : acquire ; login credentials : compromised data ; hackers : attacker ; Two years prior : time )",
"output_json": "{\"events\": [{\"event_type\": \"data breach\", \"trigger\": \"acquire\", \"arguments\": [{\"name\": \"login credentials\", \"role\": \"compromised data\"}, {\"name\": \"hackers\", \"role\": \"attacker\"}, {\"name\": \"Two years prior\", \"role\": \"time\"}]}]}",
"sentence": "Two years prior to the first Ukraine incident , hackers began attempting to acquire legitimate login credentials by hacking non-operational systems at Ukrainian utilities – systems very much like Burlington Electric ’s laptop ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( data breach : acquire ; login credentials : compromised data ; hackers : attacker ; Two years prior : time )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "data breach", "trigger": "acquire", "arguments": [{"name": "login credentials", "role": "compromised data"}, {"name": "hackers", "role": "attacker"}, {"name": "Two years prior", "role": "time"}]}]}
|
( data breach : acquire ; login credentials : compromised data ; hackers : attacker ; Two years prior : time )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nTwo years prior to the first Ukraine incident , hackers began attempting to acquire legitimate login credentials by hacking non-operational systems at Ukrainian utilities – systems very much like Burlington Electric ’s laptop .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "data breach", "trigger": "acquire", "arguments": [{"name": "login credentials", "role": "compromised data"}, {"name": "hackers", "role": "attacker"}, {"name": "Two years prior", "role": "time"}]}]}
|
EE
|
CASIE
|
train
|
[] |
{
"ground_truth": "( phishing : were sent to ; May 2014 : time ; Ukraine electric utility employees : victim ; spear - phishing emails : tool )",
"id": "3750",
"instruction": "Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return \"None\". Return the extracted roles in a structured event list.\nOption: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ) \nText: {0} \nAnswer:",
"label": "( phishing : were sent to ; May 2014 : time ; Ukraine electric utility employees : victim ; spear - phishing emails : tool )",
"output_json": "{\"events\": [{\"event_type\": \"phishing\", \"trigger\": \"were sent to\", \"arguments\": [{\"name\": \"May 2014\", \"role\": \"time\"}, {\"name\": \"Ukraine electric utility employees\", \"role\": \"victim\"}, {\"name\": \"spear - phishing emails\", \"role\": \"tool\"}]}]}",
"sentence": "According to a Booz Allen analysis , spear - phishing emails containing weaponized Microsoft Word , Excel , and PowerPoint files , exactly the type of files typically found on laptop computers , were sent to Ukraine electric utility employees as early as May 2014 ."
}
|
Identify and extract the participants (roles) involved in the described event based on the provided event type. If no valid event participants are found, return "None". Return the extracted roles in a structured event list.
|
( phishing : were sent to ; May 2014 : time ; Ukraine electric utility employees : victim ; spear - phishing emails : tool )
|
Option: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern )
|
{"events": [{"event_type": "phishing", "trigger": "were sent to", "arguments": [{"name": "May 2014", "role": "time"}, {"name": "Ukraine electric utility employees", "role": "victim"}, {"name": "spear - phishing emails", "role": "tool"}]}]}
|
( phishing : were sent to ; May 2014 : time ; Ukraine electric utility employees : victim ; spear - phishing emails : tool )
|
Extract all events from the given text and return them in JSON format with the key "events". The output should be a JSON string containing a list of objects, each with "event_type", "trigger", and "arguments" (a list of objects with "name" and "role").
|
[
{
"content": "You are an AI assistant specialized in EE tasks for the CASIE dataset. \nYour goal is to follow the instructions carefully and generate accurate outputs in JSON format.\nThe available options for this task are: Event type: ( phishing, data breach, ransom, discover vulnerability, patch vulnerability ) . Arguments type: ( vulnerable system, capabilities, vulnerability, number of victim, issues addressed, compromised data, damage amount, common vulnerabilities and exposures, victim, patch number, number of data, tool, discoverer, time, purpose, place, payment method, releaser, vulnerable system version, attacker, supported platform, patch, price, vulnerable system owner, trusted entity, attack pattern ).",
"role": "system"
},
{
"content": "### Instruction:\nExtract all events from the given text and return them in JSON format with the key \"events\". The output should be a JSON string containing a list of objects, each with \"event_type\", \"trigger\", and \"arguments\" (a list of objects with \"name\" and \"role\").\n\n### Input:\nAccording to a Booz Allen analysis , spear - phishing emails containing weaponized Microsoft Word , Excel , and PowerPoint files , exactly the type of files typically found on laptop computers , were sent to Ukraine electric utility employees as early as May 2014 .\n\n### Output:",
"role": "user"
}
] |
{"events": [{"event_type": "phishing", "trigger": "were sent to", "arguments": [{"name": "May 2014", "role": "time"}, {"name": "Ukraine electric utility employees", "role": "victim"}, {"name": "spear - phishing emails", "role": "tool"}]}]}
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.