Research On The Method Of Expressing Results Of Co-Word Clustering Analysis Based On Sentence Extraction

Objective: The expression of the results of co-word cluster analysis is the last step in the application of co-word analysis.Existing methods for expressing results of co-word analysis have certain limitations,and the results obtained have problems of strong subjectivity and incomplete expression.This study proposes a method for expressing cluster analysis results based on sentence extraction,to promote the objectivity,accuracy,ease of understanding and standardization of the expression of the results of common word analysis.Provides assistance and reference in the expression of word analysis results,and to a certain extent,promotes the further development of information analysis technology in the analysis of professional hot topics.Methods: In this study,we obtained the clustering results of the subject words at two levels and tried to express their results: one is the results of the common word analysis in the published hot analysis articles,and the other is the clustering by the researchers according to the topics in the review articles of the domain experts.The results of the co-word analysis obtained after the class analysis take these two clustering results as the research sample.For the two types of clustering results,first,based on the clustering order of the subject words in the clustering results,according to certain rules to construct a combination of subject words within the class,formulate and execute the corresponding search strategy to obtain a document set related to each subject word combination;Then,through natural language processing technology,extract the semantic relationship expression of all subject word combinations in the abstract of the document collection,and use the main predicate and the representative sentence between the two concepts as the result of the word interpretation of the cluster;finally,on the one hand,the two results obtained in this study are comparedwith the corresponding number of subject words for co-word analysis to evaluate the coverage of the subject words by this method,on the other hand,the two results are compared with the existing co-word analysis results.Compare the content with the summary topic,and finally make a comprehensive evaluation of the applicability of this method.Results: For the published results of co-word analysis,this study constructed 81 subject words into 38 subject word combinations based on certain subject word combination rules,of which 32 subject word combinations can be retrieved from the relevant literature collection and 17 subjects Word combinations can be extracted into semantic relations.The average coverage rate of the expressions of the co-word analysis results obtained by this method to the intra-topic keywords is 56%.A total of27 semantic relationship expressions have been obtained for the 17 intra-topic keyword combinations;compare them with the expression expression in the published paper It is found that the content consistency is 73%,and the interpretation of the results obtained in this article is more specific and has better readability.For the co-word analysis based on the selected review topic,9 topic words that are closely related to the research topic form 6 intra-topic keyword combinations,of which 5topic keyword combinations can extract semantic relations.The average coverage of the clustered topic expressions obtained by the method in this paper is 83%.The combination of the five topical keyword combinations yields a total of 8 semantic relationship expressions;however,the results of the co-word analysis and the content of the original review The comparison found that there is a certain degree of inconsistency between the co-word analysis results and the summary content.Conclusion: The research results show that the hotspots obtained by cluster analysis of subject words need to be further studied to meet the practical needs of experts.After coming out,the obtained interpretation of the cluster content has certain objectivity,accuracy and standardization,which provides an effective way to further standardize the expression of the results of co-word clustering.