Update README.md

README.md

@@ -203,3 +203,35 @@ configs:
   - split: train
     path: data/train-*
 ---
+
+### Dataset Description
+
+Following data collection and processing, we have created the first dataset comprising a total of 4,379 GitHub issues. Of these, 844 are linked to disclosed CVE vulnerabilities, while 3,535 represent issues unrelated to any vulnerabilities. These issues are taken from 31 different repositories. For each issue linked to a vulnerability, we include CVE ID, disclosure date, vulnerability description, a list of references, details on the affected software, and vulnerability impact metrics updated until August 11th, 2024. 
+
+- **Paper:** [Automating the Detection of Code Vulnerabilities by Analyzing GitHub Issues](https://arxiv.org/abs/2501.05258)
+
+### Fields
+
+1. **cve_id**: The unique CVE identifier.  
+2. **cve_published**: The timestamp when the CVE was published.  
+3. **cve_descriptions**: Textual descriptions of the CVE.  
+4. **cve_metrics**: Container of CVSS metric information.  
+5. **cve_references**: List of external reference links.  
+6. **cve_configurations**: Details about affected configurations.  
+7. **url**: The primary link to the source or CVE entry.  
+8. **cve_tags**: Tags categorizing the CVE.  
+9. **domain**: Domain of the refenrece link. 
+10. **issue_owner_repo**: Repository owner and name from GitHub.  
+11. **issue_body**: Body text of the GitHub issue.  
+12. **issue_title**: Title of the GitHub issue.  
+13. **issue_comments_url**: API URL for issue comments.  
+14. **issue_comments_count**: Number of comments on the issue.  
+15. **issue_created_at**: Timestamp of issue creation.  
+16. **issue_updated_at**: Timestamp of the last issue update.  
+17. **issue_html_url**: Web link to view the issue on GitHub.  
+18. **issue_github_id**: GitHub's unique ID for the issue.  
+19. **issue_number**: Issue number in the repository.  
+20. **label**: Boolean label (true/false) for classification.  
+21. **issue_msg**: Prompt used for the generation of the embedding.  
+22. **issue_msg_n_tokens**: Number of tokens in the prompt.  
+23. **issue_embedding**: Embedding of the prompt generated with text-embedding-3-large.