<$BlogRSDUrl$> Marcus P. Zillman, M.S., A.M.H.A. Author/Speaker/Consultant
Marcus P. Zillman, M.S., A.M.H.A. Author/Speaker/Consultant
Internet Happenings, Events and Sources


Sunday, January 23, 2005  

Got A Question? Talk To Your Agent
http://www.firstmonday.org/issues/issue10_1/galitsky/index.html

The role of automated advisors in e-commerce and various financial and business services has grown dramatically. Providing fast answers to questions about products and services has become an important marketing tool. Now the challenge is to build a question answering (Q/A) system with the precision of a vertical domain and the coverage of a horizontal domain. The best solution may be multiagent architecture, which allows each agent to answer questions in a vertical domain, while an entire community of agents -- i.e., the multiagent community -- covers a horizontal domain. A new report suggests the multiagent question-answering approach is adequate for natural language question-answering systems, in which the sources of knowledge are quite heterogeneous. The meta-agent controls the cooperation between question answering agents and chooses the most relevant answer(s). Multiagent implementation is tightly connected with the context of an entity that appears only in semantic-based systems. Context machinery is helpful when it yields a meaning of an expression (given a predicate and its attribute), compared to pure keywords, which have only individual meanings. Horizontal question answering systems cannot provide as deep and accurate question answering as multiagent ones, incorporating numerous domains. One example: AskJeeves has fewer than 50 answers (links) concerning tax topics. But the multiagent approach delivers more than 2,000 answers on tax and associated topics.

posted by Marcus Zillman | 4:10 AM
archives
subject tracers™