Research on the theme evolution trend of graduate supervisor evaluation system based on topic model

Objective: To excavate the hot topics and evolutionary features of the evaluation system of postgraduate tutors, and to examine the developmental orientation of the tutor evaluation system. Methods: Taking the Chinese Social Sciences Citation Index (CSSCI) included in China Knowledge Network (CNKI) as the data source, we use Latent Dirichlet Allocation (LDA) topic model to identify the text content, divide topic modules and use time series to analyze the evolution process of "topic intensity". Results: Four types of themes were refined: teacher ethics (Topic1), research and innovation ability (Topic2), talent training quality (Topic3) and teaching level (Topic4). Conclusion: The core themes show different flow trends over time, teacher ethics is at the core of the tutor evaluation system, and the themes of research and innovation ability and talent cultivation quality belong to the focus of attention in this field.