Improving QA performance through semantic reformulation

Crushing of textual information is available in electronic form on internet. As a result finding the answer to user query essential in natural Language processing, information retrieval and question answering. Semantic based question reformulation is frequently used in question answering system to retrieve answer from large document collection. The goal of this paper is to find useful and standard reformulation pattern, which can be used in our question answering (QA) system to find exact candidate answer. In this paper we used TREC-8, TREC-9, and TREC-10 collection as training set. Different types of question and corresponding answer can use from TREC collection. The QA system will automatically extract the pattern from sentences retrieved from search engine. With help of word net it will check syntactic tags, semantic relation between question and answer pair. Next weight can assigned to each extracted pattern according to is length, the distance between keyword and the level of semantic similarity between the extracted question and answer. The proposed systems vary from most former other reformulation learning system.