Bayesian probabilistic tensor factorization for malicious tweets in trending topics

Online social networks (OSNs) are widely known interactive communication tools for millions of users and their friends. But, in recent years, these tools become the platforms for executing spam messages and malwares. It is obvious that a user reacts to the message of an OSN friend than from an unknown user which makes the social spam more vulnerable and popular than conventional email. Twitter spam detection has become an attractive research topic in recent years in which the majority of earlier works focuses upon identifying malicious user accounts and based on using honeypot detection schemes. Conventional approaches uses Twitter trending topics without interaction of user activity and receiver. In order to handle this issue, this work proposes a new social interaction Tweet Trending Bayesian Probabilistic Tensor Factorization (TTBPTF) based tweet trending topics spammers´ detection in social networks that exploits social activities, spammer activity and user activity in Twitter. These interaction activities are analyzed in TTBPTF and features such as user profile features, semantic features, user activity and location based features are extracted. Then, the extracted features are represented in matrix format to analyze the results of tweet trending topics spam user with Bayesian Probabilistic learning methods. The extracted text features are analyzed using Quadratic entropy measure in which each labeled tweet is represented through natural language models. Moreover, Bayesian probabilistic learning method is applied with the goal of examining the emergent characteristics of spam in social networks. The proposed approach detects the spammers collectively based on the users´ social actions, spammer activity and social relations. It is observed that when the Bayesian Probabilistic Tensor Factorization is adjusted for the feature vector learning, the results of spammer activity and received user results are analyzed in an efficient manner.