Artificial Intelligence-Based Model for Data Security and Mitigation Against SQL Injection Attacks in Web Applications

The likelihood of online threats, specifically Structured Query Language Injection Attacks (SQLIA), has been steadily increasing. These attacks involve stealing personal information from vulnerable web application databases, thereby posing a serious security risk. The existing solutions for mitigating SQLIA do not account for other types of injection attacks, such as buffer overflow and database fingerprinting. Additionally, the algorithms used to develop these models may be biased, subjective, and judgmental. The availability of data-driven data sets and historical data for training classifiers also presents challenges, as the data sets are limited and often outdated. To address these issues and ensure data security, this study utilized a subset of Artificial Intelligence known as ML to predict and prevent SQLIA. Data mining paradigms were employed using ML to determine classes for both Malicious and Benign data. The dataset was analyzed using the Waikato Environment for Knowledge Analysis (WEKA) with six algorithms of machine learning which are Logistic Regression, Naive Bayes, Bayes Network, Instance Based Learner, Stochastic Gradient Descent as well as Sequential Minimal Optimization. Performance evaluation of the ML algorithms using a 10-fold cross-validation approach revealed that IBK is the best classifier, achieving impressive results with 99.999% accuracy, 100.00% true negative (TN), 0.01% false positive (FP), 100.00% true positive (TP), as well as 0.00% false negative (FN). The model building time was 0.06 second, with a model training time of 0.012 second in cross validation of 10-fold. The outputs show that the model has potentials for mitigating SQLIA in web application and to ensure security of databases.