Research On Releated Technologies Of Campus Object Search Engine Based On Repository

Information Retrieval is to identify the information that satisfies users' requirements with technologies like indexing, matching and so on. The conventional information retrieval started from text data retrieval. Aiming at the personalization of information retrieval, this dissertation proposes an effective method. The author uses structured entity attribute matching, which is prior to key word matching. With this method adopted, the information retrieval system could achieve better pertinence and higher precision, and thereby provide personalized information retrieval service on entity webpages in an effective and high quality.This paper provided an object search engine based on repository, which based on a Campus Object Search Engine (COSE) which was carried out by pattern recognition lab, this study focus on mainly the following aspects: 1,Repository and manually tagging approach are developed based on person and organization names as well as course names.2,Information extraction of Bupt corpus is performed adopting word segmentation, named entity recognization etc Natural Language Processing technology which based on object repository, than, we created the expansion repository by using the results of information extraction.3,Finally, we improved and optimized COSE system based on object repository and expansion repository.