Spammer detection and fake user detection using modified Naïve Bayes and decision Tree classifier

Earlier, when Online Source Networks (OSNs) were not popular among most of the people the fraud or spamming activities used to happen through fake calls or SMSs. As the Online Source Networks have become popular platforms for people to share their views, ideas, news, information etc. People’s involvement in these platforms are increasing over these platforms. We know that if some people are using these platforms to share true information or ideas, then there must be some people using these platforms to spread fake news, information or malicious messages. Commonly we can divide these two users in two category, Non-Spammers and Spammers. Non-Spammers will be linked to OSNs for true information while Spammers will be linked to fake content, URL based malicious activities, spam in trending topics and fake identification. Twitter is the main example of widely used OSN platform where most of the Legitimate as well as Spammers are present, unreasonable number of spams on tweets, trending topics disturbing the real source. In this research work a integrated approach of three machine learning algorithms namely Forest Research, Decision Trees and Naïve Bayes is being proposed. Results will show the accuracy of combinational approach is more than the classical approach.