Detection and prevention of SQLI attacks and developing compressive framework using machine learning and hybrid techniques

Abstract A web application is a software system that provides an interface to its users through browser on any operating (OS). Despite their growing popularity, security threats have become more diverse, resulting in severe damage. Malware attacks, particularly SQLI are common poorly designed applications. This vulnerability has been known for than two decades and still source of concern. Accordingly, different techniques proposed counter attacks. However, the majority them either fail cover entire scope problem. The structured query language injection (SQLI) attack among most harmful online attacks often happens when attacker(s) alter (modify), remove (delete), read, copy data from database servers. All facets security, including confidentiality, integrity, availability, can be impacted by successful attack. paper investigates forms, mechanisms, method identifying, detecting, preventing based existence SQL query. Here, we developed comprehensive framework detecting effectiveness address specific issues following essence using traditional Navies Bayes (NB), Decision Trees (DT), Support Vectors Machine (SVM), Random Forests (RF), Logistic Regression (LR), Neural Networks Based Multilayer Perceptron (MLP), hybrid approach used our study. machine learning (ML) algorithms were implemented Keras library, while classical methods Tensor Flow-Learn package. For this research work, gathered 54,306 pieces weblogs, cookies, session usage, HTTP (S) request files train test model. performance evaluation results training set metrics such as (ANN SVM) perform better accuracies precision (99.05% 99.54%), recall (99.65% 99.61%), f1-score (99.35% 99.57%), (99.20% 99.60%) respectively other ML approaches. time too high (i.e., 19.62 26.16 s respectively) NB RF. technique performs accuracy, precision, recall, f1-score, metrics, best time. Additionally, (98.87% 99.20%), (99.13% 99.47%), (99.00% 99.33%) (98.70% 99.40%) 11.76 15.33 ms respectively). techniques, SVM ANN weak learners. achieved indicated detection prevention mechanism improved over previously theme. Finally, paper, aimed keep researchers up-to-date, with contributions, recommendations understanding intersection between artificial intelligence (AI) field.