The Analysis And Design Of The Patent Retrieval System Based On Knowledge Discovery In Database

Patent technology is an important factor to reflect the science, technology innovation and latent competitiveness level of a country, a department or an enterprise. An accurate and efficient patent retrieval system is a bridge which plays an important role between patent holders and patent demanders, it can improve not only the patent conversion rate of a country, but also the technology and innovation of an enterprise. Nowadays, although the quantity of the patent in China is increasing rapidly, market demand of patent is expanding constantly, the patent conversion rate is very low. The reason is that the retrieval of existing patent retrieval systems are almost only based on the key words. In this case, the retrieval results rely on search keywords heavily, and it makes excessive demands of users. All the above cause the difficulties of patent retrieval.From on macroscopic, researching a convenient and practical patent retrieval system to solve these problems can promote implementation of patent transformation and enhance our patent level. Microscopically speaking, a suitable patent means a opportunity of development for a corporation, it is directly related to enterprise development prospects. This subject is a new idea of patent retrieval, the paper applies knowledge discovery method in the patent retrieval to construct a highly efficient and accurate patent retrieval system, providing convenient and effective retrieval methods to users.The research content of this paper is about the patent retrieval system based on Knowledge Discovery in Database. The paper analyses the deficiency of existing patent retrieval systems, and puts forward a patent retrieval system based on knowledge discovery according to the shortcomings of the existing retrieval systems, it is the integration between the whole process of knowledge discovery and patent retrieval process for realizing patent retrieval with knowledge discovery method. The patent retrieval process mainly includes three parts: data preparation, keywords expansion and calculation of similarity. In the whole process, the system uses association analysis, vector space model and similarity calculation method synthetically. Firstly, patent text documents checked initially are preprocessed through word segmentation, word filtering and feature selection. Secondly, in order to make the retrieval results more comprehensive and accurate, the system uses association analysis to expand keywords. Finally, patent text documents checked initially are ordered again after patent character representation based on space vector model and similarity calculation based on included angle cosine. The final retrieval results are presented to the user finally. Different from the traditional patent retrieval, the retrieval results through the patent retrieval system in this paper are more accurate and comprehensive. In addition, the case in chapter5of this paper proves that the patent retrieval system based on Knowledge Discovery in Database is effective and applicable.This paper uses kinds of knowledge discovery and data mining methods synthetically, proposing a new idea of patent retrieval. The research provides reference to the patent retrieval technology of our country, it is beneficial to the development of the patent retrieval technology, the patent analysis and the development of the patent in other fields.