Enhancing Vulnerability Reports with Automated and Augmented Description Summarization

Public vulnerability databases, such as the National Vulnerability Database (NVD), document vulnerabilities and facilitate threat information sharing. However, they often suffer from short descriptions and outdated or insufficient information. In this paper, we introduce Zad, a system designed to enrich NVD vulnerability descriptions by leveraging external resources. Zad consists of two pipelines: one collects and filters supplementary data using two encoders to build a detailed dataset, while the other fine-tunes a pre-trained model on this dataset to generate enriched descriptions. By addressing brevity and improving content quality, Zad produces more comprehensive and cohesive vulnerability descriptions. We evaluate Zad using standard summarization metrics and human assessments, demonstrating its effectiveness in enhancing vulnerability information.

Enriching Vulnerability Reports Through Automated and Augmented Description Summarization

Security incidents and data breaches are increasing rapidly, and only a fraction of them is being reported. Public vulnerability databases, e.g., national vulnerability database (NVD) and common vulnerability and exposure (CVE), have been leading the effort in documenting vulnerabilities and sharing them to aid defenses. Both are known for many issues, including brief vulnerability descriptions. Those descriptions play an important role in communicating the vulnerability information to security analysts in order to develop the appropriate countermeasure. Many resources provide additional information about vulnerabilities, however, they are not utilized to boost public repositories. In this paper, we devise a pipeline to augment vulnerability description through third party reference (hyperlink) scrapping. To normalize the description, we build a natural language summarization pipeline utilizing a pretrained language model that is fine-tuned using labeled instances and evaluate its performance against both human evaluation (golden standard) and computational metrics, showing initial promising results in terms of summary fluency, completeness, correctness, and understanding.