Context-sensitive text mining and belief revision for intelligent information retrieval on the web

Abstract

Autonomous information agents alleviate the information overload problem on the Internet. The AGM belief revision framework provides a rigorous formal foundation to develop adaptive information agents. The expressive power of the belief revision logic allows information seekers' changing information preferences and the underlying retrieval contexts to be captured in information agents. By exploiting the relevant retrieval contexts, information agents can proactively recommend interesting information items to their users. Contextual knowledge for information retrieval can be acquired by information agents via context-sensitive text mining. The induction power brought by context-sensitive text mining and the nonmonotonic reasoning capability offered by a belief revision system are complementary to each other. This paper illustrates a novel approach of integrating the proposed text mining method into the belief revision based adaptive information agents to improve the agents' learning autonomy and prediction power. Our initial experiments show that the symbolic adaptive information agents outperform their vector space model based counterparts.