Value Analysis And Visualization Of Scientific Papers Based On Citation Text

In order to screen high-value papers related to topics from a large number of papers,the value of the papers needs to be quantified and ranked.Many studies have improved the page link ranking algorithm Page Rank and applied it to paper ranking with some results.In conventional research,the focus is on citation relationships between papers,and little attention has been paid to the descriptions(ie,citation text)of references in the body of the paper.Existing research proves that the citation text can reflect the main content of the cited document and the information used by the citation person from the cited document.Analyzing the citation text can study the citation behavior in more detail,so that the citation network-based paper ranking results are more accurate.It is helpful for revealing the subject of the cited literature,so this article will incorporate citation text analysis into the process of calculating the value of the paper and classifying the literature according to its subject.Firstly,on the basis of incorporating the WPage Rank algorithm into the publication time factor of the paper,add citation text analysis,calculate the citation correlation between the papers by extracting the keywords in the citation text,and consider the journal impact factor and the publication time of the paper to calculate the paper.In this paper,the citation relevance and the intrinsic value of the paper are added to the WPage Rank algorithm,and an improved RVWPage Rank algorithm is proposed.Experiments show that the RVWPage Rank algorithm enables newly published highquality documents to obtain higher rankings,and makes high-quality documents in the field more easily retrieved.At the same time,it ensures the timeliness and topic concentration of the search,which weakens the "cold start" Negative effects.Secondly,for the problem that it is not easy to quickly find the research direction covered by the search results based on the ranking of the papers,this paper combines the citation text analysis technology,improved paper value calculation method,and LDA topic model to perform topic clustering for papers in different fields.Screen highvalue papers under each topic.Researchers can quickly understand the main research direction and research results of a certain field based on the topics and high-value papers under that topic.Finally,a citation-based sci-tech paper value analysis and visualization system was developed and combined with citation content analysis methods to visualize the value of the paper,citation content,and citation relationships between papers in the form of relationship diagrams.The histogram and other forms visualized the author's research status.