Software Vulnerability Detection Using Machin...

Software Vulnerability Detection Using Machine Learning Algorithms

Author Name : Hamza Ahmed Abdinoor, Mohammad Ibrahim Elshatarat

DOI: https://doi.org/10.56025/IJARESM.2024.1209241347

ABSTRACT This paper presents a machine learning based software vulnerabilities detection tool which detects the major security vulnerabilities like SQL Injection (SQLi), Cross-Site Scripting (CSS or XSS) and Remote File Inclusion (RFI). Such techniques include support vector machines (SVM), K-nearest neighbor (KNN) [7] and naïve bayes [2], besides ensemble methods of bagging and boosting to increase the accuracy and reliability of the tool [8]. Developed using Django framework, it enables the user to upload datasets, start the training process and visualize the results in real time. The feature extraction is done by Term Frequency-Inverse Document Frequency (TF-IDF) [9] and the performance is measured by basic performance indicators such as accuracy, precision, recall, and F1-score and these are shown by confusion matrices. The generic nature of the system helps in identifying the vulnerabilities in various software and the paper also analyses the areas of improvement like usage of more data sets, better models for improving the detection accuracy