Development of Thai question answering system

Abstract

© 2017 Copyright is held by the owner/author(s). Automatic question answering systems play an important role nowadays. In this paper, we developed a Thai question answering system. From an input question, the system analyzes an expected type of answer (e.g. human, place) then searches for all possible answers from source documents. In order to select the most relevance answer, a word order consistency and a relevance score are proposed to measure similarity of the question and the sentences surrounding the possible answers. Based on an experiment from 210 questions and 50 articles, the proposed system can provide a mean reciprocal answer rank of 0.657.