Research On Patent Novelty Analysis Technology Based On Keyword Extraction

The analysis of patent novelty is a necessary work in patent application.The traditional manual analysis of patent novelty usually consumes a lot of manpower and time,the analysis is very inefficient,and also the quality is difficult to guarantee.At present,there are two main problems of using computers to analyze patent novelty: One is that the comparative patent is not related to the technical characteristics of the patent to be applied for;the other is that the analysis conclusion of the novelty of the patent to be applied for is not reliable.For the first problem,in order to extract keywords related to patent subject,this paper uses a keyword extraction method based on the word network.In this method,the text is mapped into a semantic similarity network of words,the center of the set of word centrality is calculated,and the word centrality of the corresponding region of the center is taken as the semantic contribution value of words.For the second problem,in order to make the conclusion of patent novelty analysis more reliable,this paper uses a patent novelty analysis method based on sentence similarity.This method extracts the concept nodes and relationship nodes from the sentences that reveal the technical features of Patent Texts,constructs the concept map to calculate the sentence similarity between Patent Texts,and analyzes the novelty of patent through sentence similarity.The main contributions of this paper are as follows: 1: In the keyword extraction method,the centrality of the node is calculated by the degree centrality index and the intermediate centrality index of the node,and the word centrality of the corresponding region of the central point is taken as the semantic contribution value of the word,which can effectively filter the keywords with high centrality ratio but unrelated to the topic.2: It is proposed that the concept map is constructed by combining the concept nodes and relationship nodes which reveal the technical characteristics of sentences to calculate the sentence similarity,which can make the conclusion of novelty more reliable.Finally,an intelligent patent search system is designed and implemented,and the feasibility and effectiveness of the method are verified by experiments.