Survey on Link Anomaly Detection for Textual Stream in Online Social Network

Link Anomaly detection is one of the most important topics in social network. Many of the social networks such as Facebook, Google+, LinkedIn, or twitter require an effective and efficient framework to identify deviated data. Anomaly detection methods are typically implemented in social stream mode, and thus cannot be easily extended to large-scale problems without sacrificing computation where the user's link is generated dynamically (replies, mentions, and retweets). A new approach model i.e. probability model, this model to the capture normal linking behavior of a social network users, and propose to detect the trending topic from the social networks through the probability model. We collect anomaly score from the different user. And aggregated score feed to change-point analysis or change-point detection, or with burst detection, finally show that to detect trending topics only based on the reply/mention in social network posts. Our technique to collect number of real data from real time twitter account.