Toolkit for automatic analysis of chat conversations

Abstract

Text-based tools, such as instant messaging, forums or blogs enable online collaboration in communities all over the world. Data sources that are more conversational in nature tend to have a typical, time-driven structure that existing text analyzing tools don't take into consideration. In this paper we propose a toolkit for information extraction from chat sessions. The system is optimized for conversational structures, integrates natural language processing with knowledge manipulation and permits to detect the topics of chat, the threads of discussion, lexical chains, the interferences between replies and coreferences. ©2010 IEEE.