A Semantic Learning-Based SQL Injection Attack Detection Technology

• DOI: 10.3390/electronics12061344
• Publication Date: 2023-03-13T06:31:55Z
• Authors: Dongzhe Lu; Jinlong Fei; Long Liu
• Keywords: SQL Injection

Over the years, injection vulnerabilities have been at top of Open Web Application Security Project Top 10 and are one most damaging widely exploited types against web applications. Structured Query Language (SQL) attack detection remains a challenging problem due to heterogeneity loads, diversity methods, variety patterns. It has demonstrated that no single model can guarantee adequate security protect applications, it is crucial develop an efficient accurate for SQL detection. In this paper, we propose synBERT, semantic learning-based explicitly embeds sentence-level information from statements into embedding vector. The learns representations be mapped syntax tree structures, as evidenced by visualization work. We gathered wide range datasets assess classification performance results show our approach outperforms previously proposed models. Even on brand-new, untrained models, accuracy reach 90% or higher, indicating good generalization performance.