Towards a Knowledge Graph Based Approach for Vulnerable Code Weaknesses Identification

A large literature exists that investigates the use of learning models for vulnerabilities detection. Even if it is well-recognized that the presence of code weaknesses (CWEs) in code fragments can make them vulnerable, most of the effort has been spent to propose new models and methods for the early detection of vulnerable code in a dataset of code fragments. This paper reports a preliminary study conducted to adopt Knowledge Graphs (KGs) for detecting the presence of CWEs in the code. The proposed approach has been evaluated on two datasets and has shown interesting results.