Software Vulnerability Detection in Source Code Using Superb Fairy-Wren Deep Transformer Guided Model for Next Generation Software Security

Finding vulnerabilities in source code is a top priority in software development, particularly for programming languages like Java are extensively utilized in operating systems, embedded software, and mission-critical applications. However, traditional systems are continuously challenged by the changing nature of software development methodologies and the growing sophistication of cyber threats, which gradually reduces their usefulness. In the proposed work, a novel optimization-assisted artificial intelligence (AI) (OAssAI) model is proposed for source code vulnerability prediction. Initially, the proposed model encompasses preparing a dataset Java source code from OWASP benchmark test pocket. Then, the feature extraction process is carried out using Doc2Vec and Term FrequencyKeyword Active Index (TF-KAI). Accordingly, a Superb Fairywren Transformer BiLSTM (SF-TransBiLSTM) model is applied for detection. The model is assessed using performance measures like accuracy, precision, recall and f1-score. Lastly, it is evident from the simulated results that the OAssAI method performed better than the existing methods. The OAssAI method achieves a maximum accuracy of 98.2 %, which proves the superiority in vulnerability detection.