Abstract: With the rapid development of the Internet, social network has become an important social tool in daily life. However, the abnormal users in social networks emerge in an endless stream, and its harm is becoming increasingly serious. Therefore, identifying and detecting abnormal users in social networks plays an important role in improving the user experience and maintaining a good network environment. This paper introduces different types of abnormal social network users, and introduces the research progress of each type of abnormal user. Finally, it summarizes the anomaly detection methods, and divides the anomaly detection technologies in social networks into categories, clustering, statistics, information theory, hybrid and graphs, and the advantages and disadvantages of these six technologies are compared, which can help people understand abnormal users and anomaly detection technologies in social networks, and provides ideas for solving abnormal problems.

Key words: social network, abnormal user, abnormity recognition technology, machine learning

摘要： 随着互联网的迅速发展，社交网络已经成为人们日常生活中的重要社交工具。然而，社交网络中的异常用户层出不穷，其危害也日益严重。因此，识别和检测社交网络中的异常用户对提高用户体验、保持良好的网络环境等具有重要作用。介绍了不同类型的社交网络异常用户，并对每种不同类型异常用户的研究进展进行了介绍；对异常检测方法进行了综述，将社交网络中的异常检测技术分为分类、聚类、统计、信息论、混合、图六大类，并对这六类技术各自的优缺点进行了比较，有助于人们了解社交网络中的异常用户、异常检测技术，为解决异常问题提供了思路。

关键词: 社交网络, 异常用户, 异常识别技术, 机器学习