期刊名称：International Journal of Innovative Research in Computer and Communication Engineering
印刷版ISSN：2320-9798
电子版ISSN：2320-9801
出版年度：2014
卷号：2
期号：9
出版社：S&S Publications
摘要：In this paper we propose an alternative keyword search mechanism for Wikipedia, designed as a prototypesolution towards factoid questions answering. The method considers relations between articles for finding the bestmatching article. Unlike the standard Wikipedia search engine and also Google engine, which search the articles contentindependently, requiring the entire query to be satisfied by a single article, the proposed system is intended to solvequeries by employing information contained in multiple articles. Although still a keyword search, the method can befurther employed in natural language questions answering, when accompanied with a question processing module. Themethod assumes that queries are formulated in a form of a list of Wikipedia articles. The possible solutions are thenevaluated, however not by attempting to understand the meaning of the text, but by a simple method of estimating thedistance between articles by measuring articles’ references or appearances in other articles, leading finally to returninga single article as an answer for the query.
关键词：natural language processing; search engine; semi-structured text; open-domain questions
