戦後日本社会学のトピックダイナミクス：『社会学評論』の構造トピックモデル分析

There is rapid progress in the field of text analysis with the development of new methodologies such as machine learning and natural language processing in recent days and has been growing attempts to apply those methodologies to the field of history of sciences and social sciences. In this study, I attempt to analyze topic dynamics of post-war Japanese sociology with one type of new large-scale text analysis methods -a topic model. Specifically, I tackle the following problems: to what extent topic model analysis can replicate findings by previous studies and whether the analysis enables to uncover hidden topics or patterns previous studies have overlooked. The data comes from texts including not only title or abstracts but also main texts published in Japanese Sociological Review from 1951 to 2015. I use a structural topic model that allows to incorporate publishing years as covariates in order to examine topic dynamics. The results show that topic model analysis can replicate previous findings to some extents and uncover hidden topic trends. The validity of these findings is further confirmed by simpler frequency analysis.