Text mining for chat message analysis

Abstract

Instant Messaging (IM) is a peer-to-peer service for remote users to communicate with each others. There are many IM systems such as MSN Messenger and Yahoo Messenger which are used by millions of users everyday. However, IM technology serves as a double edged sword and could be misused for illegitimate information exchange or committing crimes for its anonymity and completely uncontrolled chatting environment. To help enforce legitimate contents communicated in chat environments, IM monitoring systems have been developed for monitoring chat messages. Although most of these systems can provide good monitoring functions, they only provide simple message analysis features such as browsing and simple keyword-based searching of the recorded messages. In this paper, we propose a system, called IMAnalysis, that supports intelligent chat message analysis using text mining techniques. The IMAnalysis system provides functions on chat message retrieval, social network analysis and topic analysis. Chat message retrieval provides general browsing and retrieval support. Social network analysis discovers the social interactions of IM users with their contacts. And topic analysis detects automatically the topics that IM users are involved in.