Automated query analysis techniques for semantics based question answering system

Abstract

Search engines have always played an important role in helping web users to rapidly find information on the Web. However, their function is limited to returning a list of query relevant documents with reasonably good precision, but huge recall. The task of actually processing the returned documents to get the required information is the responsibility of the user. In recent years, Question-Answer systems are gaining popularity and have garnered much research interest in view of the proposed Semantic Web and future availability of fully structured data. The advantage of QA systems is that users have the luxury of asking queries in natural language and also get a precise answer instead of just displaying a list of links to documents that may or may not be relevant. This paper presents a question answer search engine prototype that uses natural language processing, natural language generation, question classification and query logs to find a precise answer to the submitted query. This is ongoing work and we focus on the methodology of query analysis in this paper. We describe our strategy of automatic query analysis by classifying it into nine categories and understanding the meaning of the query. We also discuss in detail how each of the question categories are automatically processed and how the proposed system determines the key word or key phrase to be searched.