Open-domain question answering (QA) represents a new challenge to both commercial applications and academic research. When users have specific questions, such as "What countries did Clinton visit in 1999?" or "How much does a ThinkPad cost?", they would like to see one (or a few) succint answer(s). This workshop will focus on technical issues that directly apply to this challenge, in particular, theoretical and pragmatic issues involved in the creation, evaluation and implementation of QA techniques. We concentrate on QA that is automatic and either domain independent or working within a large open domain, such as news or technical support.
To accommodate this need for automatically finding answers to open-domain questions, several different fields of research come together - information retrieval, natural-language processing and knowledge representation. This workshop will provide a forum for discussions of QA as the combination and integration of techniques from these three fields. We invite papers that deal with QA topics or components in one or more of these fields, such as answer identification using passage retrieval techniques, linguisitc analysis of questions to determine their focus, parse-based matching of questions and answers, text generation as used for formulating answers, deriving answers from knowledge bases, defining confidence measures for answers, etc. Other QA issues of interest are evaluation methods, user interface issues and user studies, integration of QA within larger systems, and commercial applications of QA.
Topics of interest include, but are not limited to:
All papers should specifically focus on question answering.