Answering FAQs: An Intelligent Approach for Extracting Answers for Queries From Subject-Oriented Multidocuments

From the ocean of text data, Extracting the important and query based text is a big NLP problem. It is a difficult task, as it requires mining text content in shortest answer length accurately and efficiently. This paper describes an attempt toward solving this problem with an intelligent approach. The proposed idea is based on the human intelligence-once a document is seen; it takes less time to answer a query from that particular document. When system becomes familiar with the document by reading more than once, it takes less time and effort to answer i.e. system uses a self-developed artificial memory. A function σ(q, D) is defined as a sequence of steps to compute the answer Λ for query q from the given set of documents D. System takes a query and set of documents as input. On the basis of the weight of keyword term(s) in query the important sentences are obtained by calculating cohesiveness with query, and then categorized to get and order of relevence by using fuzzy membership function. The most cohesive sentences are given as possible answer options to the user, to get answer Λ. The intelligent module generates an index table of keywords with page# & Para# and marks the contextual text in document; learns the queries, corresponding cohesiveness with answers, which is used for answer prediction to the next query . KeywordsFAQ, Query, Multidocument Summarization, NLP, Case Frames, Fuzzy Logic etc.