Machine Learning Models for SQL Injection Detection: A Comparative Analysis and Hybrid Framework Proposal

Baklizi M, Alkhazaleh M, Al Sardy L (2026)

Publication Type: Conference contribution

Publication year: 2026

Publisher: Institute of Electrical and Electronics Engineers Inc.

Conference Proceedings Title: 2026 2nd International Conference on Computational Intelligence Approaches and Applications, ICCIAA 2026 - Proceedings

Event location: Amman JO

ISBN: 9798331556587

DOI: 10.1109/ICCIAA68481.2026.11543922

Abstract

SQL Injection (SQLi) has continued to be among the most severe threats to web application security and has always been featured in the OWASP Top Ten. Conventional signature-based and static analysis can not cope with new attack vectors and obfuscation. In this paper, the authors provide an in-depth comparison of machine learning (ML)-based SQLi detectors, comparing the performance of traditional and deep learning architectures and ensemble-based approaches on a curated and modern dataset. We prove that our hybrid models that combine natural language processing (NLP) with ensemble learning outperform better, and the F1-scores are more than 99.2. In addition, we introduce SQLiDetect-Hybrid, a new framework that consists of syntactic features extraction, contextual embedding with language models and a stacking ensemble. The framework has a false positive rate (FPR) of 0.42% and real-time (<15ms) queries. We find empirical support to integrate adaptive ML-based detection in the next generation web application firewalls (WAFs) and intrusion prevention system (IPS).

Authors with CRIS profile

Loui Al Sardy Computer Science 7 (Computer Networks and Communication Systems)

Involved external institutions

University of Petra (UOP) / جامعة البتراء JO Jordan (JO) Isra University PK Pakistan (PK)

How to cite

APA:

Baklizi, M., Alkhazaleh, M., & Al Sardy, L. (2026). Machine Learning Models for SQL Injection Detection: A Comparative Analysis and Hybrid Framework Proposal. In 2026 2nd International Conference on Computational Intelligence Approaches and Applications, ICCIAA 2026 - Proceedings. Amman, JO: Institute of Electrical and Electronics Engineers Inc..

MLA:

Baklizi, Mahmoud, Mohammad Alkhazaleh, and Loui Al Sardy. "Machine Learning Models for SQL Injection Detection: A Comparative Analysis and Hybrid Framework Proposal." Proceedings of the 2nd International Conference on Computational Intelligence Approaches and Applications, ICCIAA 2026, Amman Institute of Electrical and Electronics Engineers Inc., 2026.

BibTeX: Download