A Novel SQL Injection Detection Using Bi-LSTM and TF-IDF
Imam Ghozali, Misbachul Falach Asy'ari, Sulaiman Triarjo, Hanun Masitha Ramadhani, Hudan Studiawan, Ary Mazharuddin Shiddiqi · 2022
SQL injection is one of the biggest threats to websites. A survey shows more than 300,000 attacks, 24,6% are SQL injection. Detection SQL injection is a complicated task because attackers can continue to change the query structure. Traditionally, SQL injection is detected using a deny list of several keywords, which is commonly used for SQL injection. However, this method is not effective anymore because the attacks have increased significantly. This research develops a technique to detect SQL injection using TF-IDF and Bi-LSTM to produce high accuracy. Experiment results indicate that the proposed method improves SQL injection detection with accuracy, precision, recall, and F1-score reaching 0.99.