| Recently, with the development of World Wide Web (WWW), technology of search engine and information retrieval has got fast development. Due to the opening propertity, information on the Internet is getting more and more abundant.Although the internet provides immense information for users, it happens "information redundancy". Users difficultly obtain needed information from the Web. So a new technology about information retrieval--- Question- Answering System(QA System) appears.Firstly, the paper introduces the background of QA System, overseas and home present situation and the investigative content. Secondly, the paper presents the Chinese words-segmentation technology and the question analysis. Whereafter, the paper details the main works:information retrieval and answer extraction.In information retrieval, the paper introduces the Agent technology of meta search engine, got the sets of candidate abstracts and URLs from the member search engines. At last, the meta search engine Agent got the first answering set with HTML label language by deleting redundancy Web-Site and judging validity of Web-Site. Therefore, the recall rate of the cadidate answering is improved and the answering range for the answer extraction is extended widely with the meta search engine.In answer extraction, the paper introduces technology route of multi-document extraction in order to extract answer with it. The regular expression is used to clean the web pages of the first answering set, the vector space model (VSM) is used to calculate similarity between question and each answer in order to get the second answering set. After splitting sentences of the second set, the final answer is obtained by similarity calculating based on words surface layer character.The paper achieves all modules of the QA System, and implements the system testing and the evaluation analysis.At last, the paper summarizes the works and provides the further working direction. |
PDF Full Text Request