Application Of Uncertain Semantic Retrieval In Patent Intelligent Service Platform

Patent information is the crystallization of human intelligence.It contains the most important technology achievement of almost every application field such as economic development,technology innovation,and strategic decision.Nowadays,patent information has become the most necessary information resource in every country's economic development.This paper combined uncertain semantic retrieval with patent information retrieval, gave semantic meaning to patent retrieval based on traditional information retrieval,which was searched by key words,in order to improve recall and precision ratio.The main works of this paper included four points as follows:(1) Analyzed patents' data structure of famous patent retrieval system domestic and abroad and laid a strong emphasis on two problems:distributed and heterogeneous patent data and low retrieval precision.The present research and relative technology of data integration and information retrieval were summarized at the same time.(2) This paper summarized the common characters of heterogeneous patent data from different countries and organizations by the way of discussing data structure of China, seven other countries and two organizations.And then designed common data transform model to integrate patent data.Because the patent information is very large,we studied on database cluster technology and used Oracle RAC to store data which can improve the retrieval efficiency.(3) Researched on personalized retrieval requirements of different users and constructed user interest ontology.Firstly,found the user interesting key words from users' login information and visit traces,calculated the weight of key words,chose the height weight key words to form the user profile text.And then constructed user interest ontology based on user profile text.(4) Based on the technologies mentioned above,we constructed a patent information service platform and realized efficient retrieval on patent information.It was proved that the retrieval efficiency was improved after heterogeneous patent data's integration and user interesting ontology's addition.