Research on Cross-Project Vulnerability Detection Method for Open Source Software

This paper centers on applying and advancing Embedding models within open-source software for cross-project vulnerability detection. It reviews relevant studies and delves into how Embedding technology captures open-source software features from multiple angles, such as code structure, community activity, and developer behavior. By transforming high-dimensional sparse data into low-dimensional dense vectors, it enhances the accuracy and reliability of feature representation. Moreover, it comprehensively examines the strengths, limitations, and future research directions of current methods, aiming to offer comprehensive theoretical and practical guidance for the intelligent analysis and security enhancement of open-source software.