Research On Patent-oriented Technology Evolution Analysis Method

The invention and innovation of science and technology is one of the important forces to promote social and economic development.In order to obtain a leading position in the research and development of new technologies,the R & D personnel of enterprises or institutions urgently need to accurately grasp the technological development context and predict technological development trends.The feasible method is to sort out the technological evolution process and stimulate innovation points based on this,and then form a new research direction.As a carrier of technology description,patents provide important data support for the research of technological evolution.Therefore,the research on patented technology evolution analysis method is meaningful and applied value.Patent mining faces many challenges,such as related field-specific terminology,it is difficult to mine accurate technical information from patents,the complexity of natural language word formation,and the diversity of expressions.In order to meet these challenges,this article analyzes Based on the characteristics of the patent,the research method of technology-oriented relationship identification was completed,and the patent technology evolution analysis was completed in combination with TRIZ theory which is known as Teofiya Resheniya Izobretatelskikh Zadatch theory.Finally,the patent data set in the field of data mining was used to verify the effectiveness of the method.The specific research work carried out is as follows:1.Aiming at the characteristics of the content of patent documents,this paper proposes a technology-oriented relationship identification method.This method first constructs a domain dictionary based on background knowledge,and uses the domain dictionary to supervise corpus word segmentation and train word vectors to cluster candidate topic words.Generate topic terms that can represent patents.Then,combine the identified topic words with semantic labels to mark the relationship data to train the relationship recognition model.Then,use the relationship recognition model to identify the three types of relationship terms:background,scheme,and effect.Finally,use Patent data in the field of data mining to verify the effectiveness of the method.2.Aiming at the shortcoming of the subjective process of technology evolution analysis,a method of technology evolution analysis based on TRIZ theory is studied.This method requires the use of technology-oriented relationship identification methods to identify three technical relationships: background,scheme,and effect.Next,the technology development s-curve combined with TRIZ theory is used to determine the development stage of the technology.Then,the classified technical relationship is used to replace the patent document and the topic discovery model is used to discover the technical topic.Finally,the topic distribution probability is generated in combination with the patent application time the change diagram of the value.Experiments show that analyzing the technology evolution map constructed by this method is helpful to understand the details of the technology evolution process.