Research on the theme evolution of graduate dissertation quality evaluation system based on LDA topic model

Objective: To identify the hot topics and evolutionary characteristics of the graduate dissertation quality evaluation system, and to promote the high-quality development of the graduate education system. Methods: Using the Chinese Social Sciences Citation Index (CSSCI) included in CNKI as the data source, the Latent Dirichlet Allocation (LDA) topic model was used to read the text, divide the topics, and use time series to analyze the "topic intensity" flow process. Results: Four types of topics were extracted: process quality (Topic 1), sampling results (Topic 2), academic innovation (Topic 3) and expert review opinions (Topic 4). Conclusion: Hot topics show different orientations with the passage of time window, and process quality is at the core of the graduate dissertation quality evaluation system, and academic innovation and expert review opinions are the focus of attention in this field.