Research And Implementation Of Fusing Ontology And Users' Interests In Patent Information Retrieval System

Patent literatures gather informations from domains including technique, law and economy and reflect latest technology research status. Since enhancing independence and innovate ability was proposed, patent information service gets increasing attention because of showing technique developing direction. Patent retrieval as the core of patent information service proposes an effective approach to gain patent literatures.Firstly, analyze and compare some famous patent retrieval systems both here and abroad. Their intelligentization hierarchy of present extensively employed patent retrieval systems is not high. They are unable to get semantic information from the search conditions and can't express users'personalized demands. Semantic web was proposed by Tim Berners-Lee in 1998. Ontology as a modeling tool describes concepts of information term system in semantic and knowledge hierarchy is applied ever-increasinly. This paper designs the patent retrieval domain ontology based on the introducing ontology technique, then uses the ontology to express the international patent classification table and users'interests, and also implements the patent retrieval system fusing ontology and users'interests.The main works of this paper include four points as follows:(1) Patent retrieval domain ontology is designed through analyzing the common concepts in searching patent information. And the international patent classification table is expressed using ontology;(2) Design users'interests model from the information actively provided by users and passively provided by system based on users'requirements and historical searching record;(3) Design and implement three patent retrieval modes: quick search, table search, and advance search to meet various users'demands. And combine ontology, user's interest model and the inputed retrieval condition when users search for patent informations. This realizes individualized retrieval which meets the demands of users;(4) When users's search term is inaccurate, the precision will drop markedly and users will receive many disrelated informations. This paper designs the classification of results according to the IPC. Users can choose any sort to browse by their own needs.Through testing the system, this paper validates that the application of ontology and user's interest model improves the recall and the precision of patent information retrieval to a certain extent. Finally, this paper summarizes all of the work and raises further work to make the system more perfect.