Processing natural language based query and context sensitive spelling suggestion in information retrieval

This paper present several key issues about modelling and processing natural language based query in information retrieval. On the basis of the sentence-pattern based module, we propose some methodologies. Focusing on misspelling and processing on natural language query, this paper also presents the strategy on context sensitive spelling suggestion as users usually present spelling errors either by accident or the proposed term has ambiguous spelling. Based on the similar comparison between the right terms and the proposed one, the algorithm can not only tolerate misspelling, but present relevant advice. It is useful in the case when user may not know how a proper term is spelled. Furthermore, if more than two correctly spelled terms are nearly tied, it can also select more common or frequently one. According to the users´ natural language query, the proposed algorithm can present relevant results or return several suggested alternative queries with the correct spelling.