Research On Core Patent Discovery Based On Improved PageRank Algorithm

Under the background of Science and technology as the primary productive forces,Patents have become the focus of attention of countries and enterprises,and the number of patent applications has risen sharply,but the vast majority of the total value of these patents are held by a few core patents that represent the most advanced technology in the industry,so how to identify the core patent from the mass of patent documents has become an urgent challenge to be solved.The identification of core patents can reflect the overall situation of an industry or a specialized technical field,provide guidance for the technological development of enterprises,prevent patent disputes,reduce the risk of operation,and have essential practical significance to the development of enterprises and industries.This paper introduces the common core patent identification methods,which are mainly divided into three aspects: based on single index and index system,based on patent reference network,co-cited network,and Page Rank algorithm.However,The identification method based on a single index is too one-sided,the identification method based on the index system has a strong subjectivity,and the identification method based on the network is too complex.Relatively speaking,the Page Rank algorithm is simple and objective enough.Therefore,starting from the original Page Rank algorithm,this paper emphatically introduces its principle,and points out the defect of the original Page Rank algorithm using average weight(transfer probability)in the iteration process.Considering the network structure,the timeliness of the patent citation network,and the patent literature information,This paper constructs the measures of literature information similarity,quality difference among patents,and patent timeliness among patents.By creating comprehensive evaluation index to replace the average weight,Page Rank algorithm is improved to get a more reasonable ranking result.This paper comprehensively considers the information similarity of patent documents,quality difference among patents,patent timeliness to create a comprehensive evaluation index to replace the weight,so as to improve the Page Rank algorithm,and further get a more reasonable ranking results.In order to verify the improved Page Rank algorithm,this paper make an empirical analysis of the Patents in the field of MIMO technology,the experimental results show that the improved Page Rank algorithm,namely the SQT-PR algorithm,compared with the original Page Rank algorithm has faster convergence speed,higher degree of differentiation.And the improved Page Rank algorithm can also withdraw the effect of timeliness,the sorting effect is more reasonable.That reflects of this paper has important research significance and practical value.