Research On Technology Evolution And Technology Opportunity Discovery Method Based On Patent Mining

How to mine technology topics from the massive patent literature,reveal the complex evolution process of technology topics and discover potential technology opportunities is crucial to improve research efficiency and grasp the competitive opportunities in science and technology.The previous technology evolution analysis methods are mostly based on the topic heat,and the analysis of topic evolution characteristics is single;the identification of final technology opportunities in technology opportunity discovery methods still relies on expert experience and lacks standardized and reliable quantitative evaluation methods.Therefore,based on the LDA model to identify technology topics,this paper proposes a multidimensional technology evolution analysis method and a technology opportunity discovery method based on linkage prediction and GBRT.And taking sewage treatment technology as an example,patent data from 2000 to 2020 are collected for empirical study.The main research work of this paper includes the following parts:(1)In order to solve the defects of the current method of dividing the time series window in the topic evolution research,the time series window is divided based on the technology life cycle theory,and the technology topics at each stage are extracted by LDA model.For the problem of single analysis of topic evolution characteristics,the evolution analysis of innovation value,innovation subject and innovation path of technology topic is carried out from the perspective of technology innovation.By constructing a scientific knowledge map with multiple dimensions,the complex evolution process of technology topics is revealed more comprehensively.(2)In order to reveal technology opportunities from the perspectives of fine-grained,quantified and semantic,etc.we collect patent data with the same development stage that the technology is currently in to construct a dataset,identify technology topics based on LDA,and divide the technology topics contained in each patent,and calculate the co-occurrence intensity between technology topics to construct a topic co-occurrence network.The feature set is constructed by CN,PA,RA and other metrics in link prediction and semantic similarity of topics,and finally the quantitative prediction of topic co-occurrence intensity is performed based on GBRT model to discover potential technology opportunities.(3)Empirical studies on sewage treatment technology are carried out to verify the effectiveness and feasibility of the method proposed in this paper.At present,sewage treatment technology is in the primary stage of maturity.Nitrogen and phosphorus removal process,biological treatment,filtration device,and adsorption materials have high technical impact and economic value during the entire development of sewage treatment technology;The social network of sewage treatment technology presents the cooperative characteristics of the joint development of universities and enterprises;Sewage treatment engineering technology is developing towards integration,automation,intelligence and resource utilization.By constructing a co-occurrence network of technical topics at each stage,and calculating 11 similarity features based on topological structure and semantic information,the prediction results of the rapid development period and the topic co-occurrence relationship of the mature period were compared and analyzed,and the dynamic accuracy rate was 60.31%.The topic co-occurrence intensity in the mature stage of sewage treatment technology is predicted,and the results show that the application of new materials in sewage treatment technology and equipment,as well as the intelligence and integration of sewage treatment equipment may be potential future technological opportunities.Based on LDA model,this paper provides a multi-dimensional analysis method for technology evolution research through topic value,topic social network and topic evolution path.A representation form of technology opportunity based on topic co-occurrence relationship is constructed.The co-occurrence intensity is predicted by GBRT model to measure potential technology opportunities,which improves the accuracy of technology opportunity discovery.