The increasing popularity of online social networks in different domains, have made convenient platforms for people to share, communicate, and collaborate with each other, which at the same time poses significant challenges and threats as many malicious behaviors, such as bullying, planning terror attacks, stealing personal information, profile cloning, social phishing, and neighborhood attacks, and physical threats. These abnormal activities in social networks are called anomalies. As social networks become a convenient platform for such malicious activities, it is extremely important to detect these activities as accurately and early as possible to avert potential threats and ensure the safety of social networking. Many researchers are now studying the usage of these networks to detect anomalous activities. There are many anomaly detection techniques available. Some of these are developed for generic applications while others are developed for specific applications. This survey provides a comprehensive overview and discussion of recent researches on different types of anomalies and their novel categorizations based on several characteristics. It also presents a structured overview of the various state-ofthe-art methods for anomaly detection. Some challenging issues of existing state-of-the-art-methods for anomaly detection are also addressed. Finally, the paper concludes with future directions and research areas.

Keywords : Online Social Networks (OSN), Anomaly Detection, Data Mining, Network Outlier Detection.