Toward a Synergistic Vulnerability Analysis Enhanced with a Multi-Level RAG Model

This paper presents a vulnerability analysis framework that has been enhanced with a multi-level Retrieval-Augmented Generation (RAG) model. The principal objective of this research is to develop a framework that will facilitate a synergistic vulnerability analysis enhanced with a multi-level RAG model. To enhance the analysis of Common Vulnerabilities and Exposures (CVE), the model progressively integrates multiple vulnerability databases, including the National Vulnerability Database (NVD), CERT Vulnerability Notes Database (VNDB), GitHub Advisory Database, and Exploit Database. By employing the RAG framework, the model retrieves pertinent information from these diverse sources and augments the analysis with accurate and up-to-date data, effectively addressing the limitations of existing fine-tuned language models. A comprehensive discussion about challenges and future works based on the model design was provided, thereby contributing valuable insights to the fields of cybersecurity and vulnerability analysis.