jinmang2 commited on
Commit
425bfe0
1 Parent(s): 62de6a2

Create r4.2/readme.txt

Browse files
Files changed (1) hide show
  1. r4.2/readme.txt +113 -0
r4.2/readme.txt ADDED
@@ -0,0 +1,113 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Release 4, Dataset 2 Notes
2
+
3
+ Major Changes
4
+ * Content is integrated with the graph structure.
5
+ * A user's topics of interest can drift over time.
6
+ * Email now includes CC/BCC.
7
+ * Email table now includes user ID and PC.
8
+ * Users can have one or more non-work email addresses.
9
+ * A latent job satisfaction variable was added. It might make sense for us to specify exactly how this affects observable variables, so let us know if that information is desired.
10
+ * An additional red team scenario was added. (All previous red team scnearios also occur in the dataset.)
11
+ * This is a "dense needles" dataset. There is an unrealistically high amount of red team data interspersed.
12
+
13
+
14
+ license.txt
15
+ * ExactData license information
16
+
17
+
18
+ logon.csv
19
+ * Fields: id, date, user, pc, activity (Logon/Logoff)
20
+ * Weekends and statutory holidays (but not personal vacations) are included as days when fewer people work.
21
+ * No user may log onto a machine where another user is already logged on, unless the first user has locked the screen.
22
+ * Logoff requires preceding logon
23
+ * A small number of daily logons are intentionally not recorded to simulate dirty data.
24
+ * Some logons occur after-hours
25
+ - After-hours logins and after-hours thumb drive usage are intended to be significant.
26
+ * Logons precede other PC activity
27
+ * Screen unlocks are recorded as logons. Screen locks are not recorded.
28
+ * Any particular user’s average habits persist day-to-day
29
+ - Start time (+ a small amount of variance)
30
+ - Length of work day (+ a small amount of variance)
31
+ - After-hours work: some users will logon after-hours, most will not
32
+ * Some employees leave the organization: no new logon activity from the default start time on the day of termination
33
+ * 1k users, each with an assigned PC
34
+ * 100 shared machines used by some of the users in addition to their assigned PC. These are shared in the sense of a computer lab, not in the sense of a Unix server or Windows Terminal Server.
35
+ * Systems administrators with global access privileges are identified by job role "ITAdmin".
36
+ * Some users log into another user's dedicated machine from time to time.
37
+
38
+
39
+ device.csv
40
+ * Fields: id, date, user, pc, activity (connect/disconnect)
41
+ * Some users use a thumb drive
42
+ * Some connect events may be missing disconnect events, because users can power down machine before removing drive
43
+ * Users are assigned a normal/average number of thumb drive uses per day. Deviations from a user's normal usage can be considered significant.
44
+
45
+
46
+ http.csv
47
+ * Fields: id, date, user, pc, url, content
48
+ * Has modular/community structure, but is not correlated with social/email graph.
49
+ * Domain names have been expanded to full URLs with paths.
50
+ * Words in the URL are usually related to the topic of the web page.
51
+ * Content consists of a space-separated list of content keywords.
52
+ * Each web page can contain multiple topics.
53
+ * WARNING: Most of the domain names are randomly generated, so some may point to malicious websites. Please exercise caution if visiting any of them.
54
+
55
+
56
+ email.csv
57
+ * Fields: id, date, user, pc, to, cc, bcc, from, size, attachment_count, content
58
+ * Driven by underlying friendship and organizational graphs.
59
+ * Role (from LDAP) drives the amount of email a user sends per day.
60
+ * The vast majority of edges (sender/recipient pairs) are exist because the two users are friends.
61
+ * A small number of edges are introduced as noise. A small percentage of the time, a user will email someone randomly.
62
+ * Emails can have multiple recipients
63
+ * Emails can have a mix of employees and non-employees in dist list
64
+ * Non employees use a non-DTAA email addresses; employees use a DTAA email address
65
+ * Terminated employees remain in the population, and thus are eligible to be contacted as non-employees
66
+ * A friendship graph edge is not implied between the multiple recipients of an email.
67
+ * Unlike the previous release, we do not believe the observed email graph follows graph power laws
68
+ because the power-law-conforming friendship graph is overwhelmed by the organizational graph.
69
+ * Email size and attachment count are not correlated with each other.
70
+ * Email size refers to the number of bytes in the message, not including attachments.
71
+ * Content consists of a space-separated list of content keywords.
72
+ * "Content" does not specifically refer to the subject or body. We have not made that distinction.
73
+ * Each message can contain multiple topics.
74
+ * Message topics are chosen based on both sender and recipient topic affinities.
75
+
76
+
77
+ file.csv
78
+ Fields: id, date, user, pc, filename, content
79
+ * Each entry represents a file copy to a removable media device.
80
+ * Content consists of a hexadecimal encoded file header followed by a space-separated list of content keywords
81
+ * Each file can contain multiple topics.
82
+ * File header correlates with filename extension.
83
+ * The file header is the same for all MS Office file types.
84
+ * Each user has a normal number of file copies per day. Deviation from normal can be considered a significant indicator.
85
+
86
+
87
+ psychometric.csv
88
+ * Fields: employee_name, user_id, O, C, E, A, N
89
+ * Big 5 psychometric score
90
+ * See http://en.wikipedia.org/wiki/Big_Five_personality_traits for the definitions of O, C, E, A, N ("Big 5").
91
+ * Extroversion score drives the number of connections a user has in the friendship graph.
92
+ * Conscientiousness score drives late work arrivals.
93
+ * This information would be latent in a real deployment, but is offered here in case it is helpful.
94
+ * A latent job satisfaction variable drives some behaviors.
95
+
96
+ Malicious actors
97
+ * This data contains two instances of insider threats.
98
+ * Data dimensions that are fair game for anomaly detection (not all are used in red team scenarios)
99
+ - In general, radical changes in behavior
100
+ - Unusual logon times (for that user)
101
+ - Unusual logins to another user's dedicated machine (for users that don't do this normally)
102
+ - Device usage for users who aren't normally device users, or increased device usage for those that are.
103
+ - Radical increases in the amount of device usage by a user
104
+ - Employee termination (as an indicator, but not anomaly detection per se)
105
+ - Number of emails sent / day
106
+ - Change in web browsing habits (visits to unusual websites are interesting, but also common)
107
+ - Radical change in social graph behavior (unexpected email recipients, perhaps)
108
+ - Topics of web sites visited, emails, and files copied.
109
+ * We can reveal as much as you would like about the red team scenarios.
110
+ * This is a "dense needles" dataset. There is an unrealistically high amount of red team data interspersed.
111
+
112
+ Errata:
113
+ * Field Ids are unique within a csv file (logon.csv, device.csv) but may not be globally unique.