SQL Detection using SVM Model over Cloud Hosted Websites
Abstract
Abstract
SQL injection (SQLi) has been constant threat to the promising businesses
shifting to online and cloud based platforms. SQLi surged since the dynamic
websites and JavaScript usage was introduced into the web development
paradigm. The hackers input malicious code using any of the input methods i.e; URL or input forms, and these malicious queries provision unauthorized access of database which can cause severe loss. To protect against these
attacks and shift our security framework from traditional manual attack detection methods to the self-learning based framework, in this research work,
we are proposing a supervised Machine Learning (ML) based Support Vector
Machine (SVM) model for the Cloud hosted websites (CSVM) to analyse and
classify web requests. We aided our model with efficient preprocessing using
No Where Clause identification and Word2Vec method for tuning of parameters. After feature extraction, CSVM is trained over an exhaustive and latest
dataset. The experimental results show improved accuracy of 98.8% SQLi attacks detection and outperforming the counterpart with significant margin.
Index Terms: CSVM, SQLi, ML , Cloud based websites, Classification algorithm