Technology-effect Matrices Constructed by Using Subject-Action-Object Semantic Analysis
Aiming at the key problem of accurately defining technology and effect themes,a method of constructing technology-effect matrices based on subject-action-object semantic analysis was proposed by analyzing semantic co-occurrence of technology and effect themes in constructing technology-effect matrices.Formulate search expressions on the basis of target domains,relevant patent data were downloaded from the patent database of China National Intellectual Property Administration,and the relevant patent data were preprocessed to obtain target patent information documents.Subject-action-object semantic structures of patent information documents were extracted by using Python language pro-gramming and Chinese word segmentation toolkit language technology platform.Semantic similarities between theme words were calculated by combining target domain corpus,term frequency-inverse document frequency,and cosine similarity.The clustering algorithm Louvain algorithm was utilized to divide community networks in order to condense the technology and effect themes,and the technology-effect matrices were constructed through co-occurrence relationship of subject-action-object semantic structures.Taking the submarine cable anti-stress cone cutting technology as an example,the effec-tiveness of the proposed method was verified through patent case analysis.The results show that the proposed method can effec-tively divide the subject-action-object semantic structures of patent cases into community networks when a large number of patent cases are analyzed to construct technology-effect matrices.By analyzing weight values of node topics in community networks to determine themes of community networks,accuracy of theme cohesion is improved.In the patent case analysis of submarine cable anti-stress cone cutting technology,7 technology themes and 9 effect themes are condensed by using subject-action-object semantic structures and Louvain algorithm,verifying the effectiveness of the proposed method.
万方数据
维普
