Research And Implementation Of Intelligent Question Answering Technology Based On Web

This paper proposes the Intelligent Question Answering Technology Based on Web (IQATBW) based on the use of answers set similarity calculation methods and answers reorganization. IQATBW has three main parts: question analysis, information retrieval and answer matching and extraction.In the module of question analysis, the lexical analysis, and syntactic analysis of Natural Language Processing (NLP) technique are used to Pretreatment the question which is questioned by the users. This process includes questions segmentation, word-tagging annotation, not-used words removal, questions classification, questions with feature items and questions with feature items extension. In this paper, the learning method of transformation-based and error-driven is used to word-tagging annotation and information gain method is used to classify the classification of question. In the process of the keyword extraction, Knowledge points,words of the questions intention and question classification words will be extracted by this system.In the module of question information retrieval, Cross-Web technology is used to build the document warehouse and Full-Text Information Retrieval (FTIR) is used to resolve answer of question information retrieval. This technique will resolve the defects of slow retrieval speed and low cross–platform property. After the information retrieval, this system will sort and page the results of questions information retrieval.In the module of the answer matching and extraction, after the FTIR will generate lots of answering fragments, a number of the front of the answering pieces can be chose to make further filter of the answer fragments. Then the collection of the answering fragments separately make similarity calculation, selecting one of the most similar answers as the final answer, and then send the final answer to the user.Lastly, in this paper, JSP technique has been realized a IQATBW system, The experiment achieved good results, but there is some room for improvement.