A survey on graph-based anomaly detection methods in social networks

The use of social networks to communicate and share information has grown dramatically in recent years. These networks are nowadays used in most areas such as education, business, health and entertainment. The large amount of valuable information on social networks has made them the main target of malicious users, such as spammers and fraudsters, for carrying out abusive and illegal activities. The abnormal and unexpected behavior of these users is identified using anomaly detection methods. Detection of anomalies is important in preventing fraud, dissemination of counterfeit information and configuration of attacks in these networks.  Anomalies are static or dynamic, with or without attributes. In this paper, various methods developed for anomaly detection in social networks have been investigated and categorized and an overview provided on anomaly detection, its applications, existing challenges and key areas for future research.