Cybercrime detection techniques based on support vector machines

Abstract

This paper presents the cybercrime detection model by using support vector machines (SVMs) to classify social network (Facebook) dataset. We try to compare between three kinds of classification algorithms such as: SVMs, AdaBoostM1, and NaiveBayes in order to find a high percentage of classification accuracy. Finally, we conclude SVMs as the best classification algorithm, which uses different breeds of kernel functions in order to improve the classification accuracy on Facebook dataset. Besides, we are using the Weka 3.7.4 software to evaluate classifiers on Facebook dataset.