Detecting Security Patches via Behavioral Data in Code Repositories
Published in arXiv / AAAI AICS, 2023
Overview
This paper explores whether security patches can be identified from developer behavior rather than source code. By analyzing commit history, collaboration patterns, and metadata across open-source Git repositories, the model classifies security fixes without inspecting the code changes directly.
Read Paper Code Dataset Google Scholar
Recommended citation: Nitzan Farhi, Noam Koenigstein, Yuval Shavitt, "Detecting Security Patches via Behavioral Data in Code Repositories." arXiv preprint arXiv:2302.02112 / AAAI Workshop on Artificial Intelligence for Cyber Security (AICS), 2023.
Download Paper