Paper Title
Rare Topic Discovery and User Behavior Analysis on Document Streams in Social Media

Abstract
Internet contains document streams that are published in various forms like posts in social media, news streams, chats etc. Among these, documents published in social media get more focus. People use social media to express their opinion about various events. These document streams are based on some topic. Many people can talk on same topic. Therefore sequential topics can be obtained from these documents. These topics are related to some rare social events, which can happen on a particular location. Also these topics can characterize user behavior. The proposed system is a text mining approach which analyses text data from social media and discover topics related to rare events. It then analyses user’s behavior towards the topic. The system contains four modules: data collection, data preprocessing, rare topic discovery and user behavior analysis. The experiment is done on twitter data which show that this approach is useful in twitter like social media sites itself. Keywords - Document Streams, Rare Events, Social Media, Topic Discovery, User Behavior.