Publication

Abstract : Web applications are the source of information suchas usernames, passwords, personally identifiable information,etc., they act as platforms of knowledge, resource sharing, digitaltransactions, digital ledgers, etc., and have been a target forattackers. In recent years reports say that there is a spike in theattacks on web applications, especially attacks like SQL injectionand Cross Site Scripting have grown in drastic numbers due todiscovery of new vulnerabilities. The attacks on web applicationsstill persist due to the nature of attack payloads, as these payloadsare highly heterogeneous and look very similar to regular texteven web applications with many security features in place mayfail to detect these malicious payload strings. To overcome thisproblem there are various methods described one such method isutilizing machine learning models to detect malicious strings byclassifying the input strings given to the web applications. Thispaper describes the study of six binary classification methodsLogistic regression, Naïve Bayes, SGD, ADABoost, RandomForrest, Decision trees using our own dataset and feature set.