Bot Detection on Online Social Network

Abstract

Social network has emerged as an important medium of communication in recent years. People join different social networks and believe the information present on it. However, these social networks face problem of bots in the system, bots are fake identities that can generate fake followings, likes, comments, posts etc. In recent years researchers have approached this problem from two basic dimensions that is content based analysis and the network traffic analysis of the social networks. In this thesis we have adopted the traffic analysis technique for bot detection. We captured and developed dataset from the traffic generated by Facebook accounts. The data is processed by our proposed model, through the modules of Processing, Aggregation, Fingerprint Generation, Subdivision and SVM. We have tested the proposed model in two different scenarios. The accuracy of the proposed model is 97%.