A Method of Author Weight Analysis for Academic Papers Based on Cluster Analysis

Abstract

The present invention relates to weight analysis of authors of academic papers, and more specifically, to a weight analysis method for authors of academic papers based on cluster analysis. In the scoring algorithm of the present invention, the first author’s score is added to the author’s citation relationship network G, and node2vec is used to model the author’s citation relationship network G to obtain the author’s word vector expression, and then use the citation relationship and the cosine similarity between nodes to score academics. Weighting is carried out to obtain the final score; after experimental demonstration, it is found that the scoring index obtained in this paper has a high correlation with h-index, and at the same time, it can weaken some defects of h-index to a certain extent; and then express according to the author's word vector Clustering is performed to obtain the category to which the author belongs and the center of each category, that is, to divide the author into different subdivisions. Relative scores within each domain are calculated by comparing the author's vector with the distance to each cluster center.

Description

technical field [0001] The present invention relates to weight analysis of authors of academic papers, and more specifically, to a weight analysis method for authors of academic papers based on cluster analysis. Background technique [0002] At present, the h-index algorithm is mainly used to evaluate the academic level of scholars. The success of h-index lies in a basic assumption in the construction of the index, that is, the level of papers can reflect the academic level of scholars to a large extent, and the papers published by scholars Quality should be more important than quantity of papers it publishes. The success of h-index is based on such insights, but due to its relatively simple design, the method has some natural defects, such as insensitivity to time variables, unable to detect the activeness of scholars over time. degree; it is not sensitive to the author’s signature order in the paper, and it is impossible to distinguish the author’s contribution in a paper...

AI Technical Summary

Problems solved by technology

The success of h-index is based on such insights, but due to its relatively simple design, the method has some natural defects, such as insensitivity to time variables, unable to detect the activeness of scholars over time. degree; it is not sensitive to the author’s signature order in the paper, and it is impossible to distinguish the author’s contribution in a paper (this is of great significance to the academic weight evaluation of scholars); the scoring values ​​​​for different fields cannot be unified, and often appear in the Scholars in one field generally have high h-index values, and scholars in one field generally have low h-index values, which will cause great errors in the evaluation of cross-field authors

Images