Integrated approach to detect spam in social media networks using hybrid features

Abstract

Online social networking sites are becoming more popular amongst Internet users. The Internet users spend some amount of time on popular social networking sites like Facebook, Twitter and LinkedIn etc. Online social networks are considered to be much useful tool to the society used by Internet lovers to communicate and transmit information. These social networking platforms are useful to share information, opinions and ideas, make new friends, and create new friend groups. Social networking sites provide large amount of technical information to the users. This large amount of information in social networking sites attracts cyber criminals to misuse these sites information. These users create their own accounts and spread vulnerable information to the genuine users. This information may be advertising some product, send some malicious links etc to disturb the natural users on social sites. Spammer detection is a major problem now days in social networking sites. Previous spam detection techniques use different set of features to classify spam and non spam users. In this paper we proposed a hybrid approach which uses content based and user based features for identification of spam on Twitter network. In this hybrid approach we used decision tree induction algorithm and Bayesian network algorithm to construct a classification model. We have analysed the proposed technique on twitter dataset. Our analysis shows that our proposed methodology is better than some other existing techniques.