The burgeoning use of Web 2.0-powered social media in recent years has inspired numerous studies on the content and composition of online social networks (OSNs)
Traditional classification algorithms perform not very well on imbalanced data sets and small sample size. To deal with the problem, a novel method is proposed