Question Answering on the Semantic Web

Question answering on the Web is moving beyond the stage where users simply type a query and retrieve a ranked ordering of appropriate Web pages. Users and analysts want targeted answers to their questions without extraneous information. These answers might contain information from current and authoritative sources, terms with the same meaning as those used in the query, relevant links such as justifications, follow-up questions fitting the context, and provenance information. Next-generation question-answering systems might also provide better querying support. This could include identifying whether questions are incoherent and therefore can't be answered, too general and would retrieve too many answers, or over constrained and would retrieve few if any answers. We present a spectrum of techniques for improving question answering and discuss their potential uses and impact.