Spammer detection based on comprehensive features in Sina Microblog
Shanshan Gao, Xiujuan Ma, Lidong Wang, Yan Yu · 2016
The popularity and accessibility of Sina Microblog have attracted a large number of spammers to conduct spamming behaviors. They can spread advertisements, disseminate pornography, virus and expose phishing. All of these behaviors are extremely harmful to legitimate users. It is necessary to detect spammers from legitimate users in Sina Microblog. In this paper, we defined two kinds of spammers, advertising spammers and following spammers. We extracted six features to distinguish advertising spammers, following spammers and legitimate users. To verify the effectiveness of our method, we use four kind machine learning algorithms, including Bayes Network, Naive Bayes, SVM and Random Forest to evaluate the spammer detection performance on our dataset. The results of these experiments show that most of the classifiers can achieve more than 90% precision rate, recall rate and F-measure rate. Bayes Network classifier was proved to the best on our dataset.