Mining Chat-room Conversations for Social and Semantic Interactions

The ephemeral nature of human communication via networks today poses interesting and challenging problems for information technologists. The Intelink intelligence network, for example, has a need to monitor chat-room conversations to ensure the integrity of sensitive data being transmitted via the network. However, the sheer volume of communication in venues such as email, newsgroups, and chat precludes manual techniques of information management. It has been estimated that over 430 million instant messages, for example, are exchanged each day on the America Online network [3]. Although a not insignificant fraction of such data may be temporarily archived (e.g., newsgroups), no systematic mechanisms exist for accumulating these artifacts of communication in a form that lends itself to the construction of models of semantics [12]. In essence, dynamic techniques of analysis are needed if textual data of this nature is to be effectively mined. This article reports our progress in developing a text mining tool for analysis of chat-room conversations. Central to our efforts is the development of functionality to answer questions such as "What topics are being discussed in a chat-room?", "Who is discussing which topics?" and "Who is interacting with whom?" The objective of our research is to develop technology that can automatically identify such patterns of interaction in both social and semantic terms. In this article we report our preliminary findings in identifying threads of conversation in multi-topic, multi-person chat-rooms. We have achieved promising results in terms of precision and recall by employing pattern recognition techniques based on finite state automata. We also report the design of our approach to building models of social and semantic interactions based on our HDDI text mining infrastructure [13].