A Restricted Domain Text Retrieval System

As appearance and popularization of Internet, people's information sources are greatlyenriched. The method of gaining information is also changed. Internet becomes one of the mainsources of gaining information. How to retrieval interesting information rapidly become anattractive research area with the exponential growth of information on the Web. Search engine wasintroduced to solve this problem.Search engine is the combination of traditional information retrieval technology and Web. Atthe initial stage of Internet development, there are relatively less websites, thus informationsearch is comparatively easy. In the instance of information scale ceaseless enlarging, facingdistributed, dynamic and large volume data, traditional information retrieval technology isincapable of finding information needful for users rapidly. Accordingly specialty search website offacing domain emerges as the times require. Facing domain search engine provides valuableinformation and correlative service aiming at certain special domain, crowd or demand. Thissystem does research on restricted domain text retrieval, and provides complete, exact andcorrelative information aiming at queries put forward by the domanial users.This system studies information retrieval which based on Vector Model, Language Model andDependency Language Model, and then from them chooses the best retrieval model. Experimentdata's comparison educes the final conclusion. Combining semantic analysis's dependency syntaxwith text retrieval based on Dependency Language Model can improve system's retrieval effect to agreat extent. After the first time retrieval the system adopts the method of query extension based onuser behavior mining. System's query extension algorithm based on former search log analysis ofusers. That is repetitious feedback result's accumulating when numerous users use retrieval system.In this way we improve the recall of text retrieval. In a word, in this paper we discussed thecorrelative technique of text retrieval and put forward how to establish an effective retrieval modelunder restricted domain. Combining text classification and expanding keywords guarantee thesystem to provide the proper information to user's queries and save mass time and vigor for users.