Question answering system for factoid based question

Abstract

Objective of question answering system (QA) is to generate concise answer of arbitrary question asked in natural language. This kind of information retrieval is required with growth of digital information. Analysis of natural language is complex task. Previously QAS were developed for specific domain and have limited efficiency. Present QAS Target on types of question commonly asked by users, characteristics of data source and correct answer generated. Our aim is to build web scale QA system. Early QA system are based on rewriting various kind of rules and pattern-generation methods for finding answer paragraph and for answer extraction. Most of QA system before answer extraction do question classification for predicting entity type of answer of question. In this paper we deals with open-domain factoid based question. Li and Roth (2002) classify questions into 6 coarse classes and 50 fine classes but it deals with limited classes of question. But we have classified question into 5 categories only and use the advance search engine technology and growth of web for QA system.