VulnMiner: A Comprehensive Framework for Vulnerability Collection from C/C++ Source Code Projects

<p>In this repository, we present an initial release of the VulnMiner vulnerability dataset, curated from prevalent projects and annotated with vulnerable and benign instances. This dataset incorporates projects with vulnerabilities labeled as Common Weakness Enumeration (CWE) categories. The developed open-source extraction tool collects vulnerability data utilizing static security analyzers.&nbsp;The study also fosters the machine learning (ML) and natural language processing (NLP) model's effectiveness in accurately classifying vulnerabilities, evidenced by its identification of numerous weaknesses in open-source projects.</p>