Main Topics and their Transitions in Over 20 Years of Self-Compassion Research: A Scoping Review Using Natural Language Processing

Objectives: Self-compassion, which refers to compassion directed toward oneself in difficult situations, is being researched in various fields, and the number of studies is increasing. However, few comprehensive scoping reviews on this exist because manually reviewing all the literature is impractical, given the large number of studies. Therefore, this study reviewed the literature on self-compassion using a computer-assisted natural language processing technique called the structural topic model. Methods: We searched for studies on self-compassion from 2003 (the year the concept of self-compassion was first defined in psychology) to July 1, 2024, using the Web of Science Core Collection platform. We applied a structural topic model to the titles and abstracts of 4,395 retrieved papers. Results: In total, 38 nameable topics were identified. Eight clusters were extracted from the network graph, showing the correlation between topics. The clusters included Various Interventions, Distress & Cognition, Eating & Body Image, Mental Health & Well-being, Childhood & Parenting, Caregiving, Sport, and Methodologies. The relationship between topic proportion, year of publication, and academic field of the journal in which it was published was visualized. Conclusions: Using topic modeling, we identified the main topics in previous self-compassion research. Furthermore, connections between topics and transitions in topics from the dawn of research to the present day were revealed. The results provide researchers with an overview of the main topics and history of self-compassion research over the past 20 years.