Datasets:

Modalities:
Tabular
Text
Formats:
parquet
Languages:
English
ArXiv:
DOI:
License:
cedricbonhomme commited on
Commit
9a55d7f
·
verified ·
1 Parent(s): e6be89b

Update dataset card

Browse files
Files changed (1) hide show
  1. README.md +8 -46
README.md CHANGED
@@ -11,49 +11,11 @@ tags:
11
  - security
12
  - cve
13
  - cvss
14
- configs:
15
- - config_name: default
16
- data_files:
17
- - split: train
18
- path: data/train-*
19
- - split: test
20
- path: data/test-*
21
- dataset_info:
22
- features:
23
- - name: id
24
- dtype: string
25
- - name: title
26
- dtype: string
27
- - name: description
28
- dtype: string
29
- - name: cpes
30
- list: string
31
- - name: cvss_v4_0
32
- dtype: float64
33
- - name: cvss_v3_1
34
- dtype: float64
35
- - name: cvss_v3_0
36
- dtype: float64
37
- - name: cvss_v2_0
38
- dtype: float64
39
- - name: patch_commit_url
40
- dtype: string
41
- - name: source
42
- dtype: string
43
- splits:
44
- - name: train
45
- num_bytes: 429220137
46
- num_examples: 641588
47
- - name: test
48
- num_bytes: 47691423
49
- num_examples: 71288
50
- download_size: 305576246
51
- dataset_size: 476911560
52
  ---
53
 
54
  # vulnerability-scores
55
 
56
- This dataset comprises **709,321** real-world vulnerabilities used to train and evaluate VLAI,
57
  a transformer-based model designed to predict software vulnerability severity levels directly from text descriptions,
58
  enabling faster and more consistent triage.
59
 
@@ -63,12 +25,12 @@ The dataset is presented in the paper [VLAI: A RoBERTa-Based Model for Automated
63
 
64
  | Source | Label | Entries | Share |
65
  |--------|-------|---------|-------|
66
- | `cvelistv5` | CVE Program (enriched with vulnrichment and Fraunhofer FKIE) | 335,972 | 47.4% |
67
- | `github` | GitHub Security Advisories | 335,376 | 47.3% |
68
- | `csaf_redhat` | CSAF Red Hat | 24,533 | 3.5% |
69
- | `csaf_cisa` | CSAF CISA | 5,860 | 0.8% |
70
  | `csaf_cisco` | CSAF Cisco | 3,930 | 0.6% |
71
- | `pysec` | PySec advisories | 3,650 | 0.5% |
72
 
73
  Extracted from the database of [Vulnerability-Lookup](https://vulnerability.circl.lu) with the [VulnTrain](https://github.com/vulnerability-lookup/VulnTrain) project.
74
  Dumps of the data are available [here](https://vulnerability.circl.lu/dumps/).
@@ -77,8 +39,8 @@ Dumps of the data are available [here](https://vulnerability.circl.lu/dumps/).
77
 
78
  | Split | Examples |
79
  |-------|----------|
80
- | train | 638,388 |
81
- | test | 70,933 |
82
 
83
  ## Fields
84
 
 
11
  - security
12
  - cve
13
  - cvss
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
14
  ---
15
 
16
  # vulnerability-scores
17
 
18
+ This dataset comprises **712,876** real-world vulnerabilities used to train and evaluate VLAI,
19
  a transformer-based model designed to predict software vulnerability severity levels directly from text descriptions,
20
  enabling faster and more consistent triage.
21
 
 
25
 
26
  | Source | Label | Entries | Share |
27
  |--------|-------|---------|-------|
28
+ | `cvelistv5` | CVE Program (enriched with vulnrichment and Fraunhofer FKIE) | 337,454 | 47.3% |
29
+ | `github` | GitHub Security Advisories | 336,720 | 47.2% |
30
+ | `csaf_redhat` | CSAF Red Hat | 25,237 | 3.5% |
31
+ | `csaf_cisa` | CSAF CISA | 5,870 | 0.8% |
32
  | `csaf_cisco` | CSAF Cisco | 3,930 | 0.6% |
33
+ | `pysec` | PySec advisories | 3,665 | 0.5% |
34
 
35
  Extracted from the database of [Vulnerability-Lookup](https://vulnerability.circl.lu) with the [VulnTrain](https://github.com/vulnerability-lookup/VulnTrain) project.
36
  Dumps of the data are available [here](https://vulnerability.circl.lu/dumps/).
 
39
 
40
  | Split | Examples |
41
  |-------|----------|
42
+ | train | 641,588 |
43
+ | test | 71,288 |
44
 
45
  ## Fields
46