SAFE: A Novel Approach For Software Vulnerability Detection from Enhancing The Capability of Large Language Models

Software vulnerabilities (SVs) have emerged as a prevalent and crucial concern for safety-critical systems. This has spurred significant advancements in utilizing AI-based methods, including machine learning and deep learning, for software vulnerability detection (SVD). While AI-based methods have shown promising performance in SVD, their effectiveness on real-world, complex, and diverse source code datasets remains limited in practice. To tackle this challenge, in this paper, we propose a novel framework that enhances the capability of large language models to learn and utilize semantic and syntactic relationships from source code data for SVD. As a result, our proposed SAFE approach can enable the acquisition of fundamental knowledge from source code data while adeptly utilizing crucial relationships, i.e., semantic and syntactic associations, to improve the effectiveness of solving the SVD problem. The rigorous and extensive experimental results on three real-world challenging datasets (i.e., Devign, ReVeal, and D2A) demonstrate the superiority of our approach over eight effective and state-of-the-art baselines. In summary, on average, our SAFE approach achieves higher performances from 4.79% to 11.57% for F1-measure and from 16.93% to 26.24% for Recall compared to the baseline methods across all the datasets used.