구조적 토픽 모델링(Structural Topic Modeling) 기반 국외 스포츠 인공지능 연구동향 분석

초록

The purpose of this study was to analyze the trends in international sports artificial intelligence research using the Structural Topic Modeling (STM) technique. The study analyzed a total of 204 papers related to sports and AI published from 1991 to 2024, collected from the Web of Science and Scopus databases. The research process comprised four steps: word preprocessing, setting the number of topics, extracting topics and setting topic names, and estimating the relationship between metadata and topic prevalence. The main findings are as follows. First, a total of 11 main topics were derived. The most frequently occurring topics were 'Basketball Strategies and Training Innovations', 'Motion Detection and Image Recognition in Sports Training', and 'Teaching and Improving Physical Education', indicating that these are traditional and comprehensive research topics. On the other hand, topics such as 'Athlete Rehabilitation and Performance Metrics' and 'Data-Driven Coaching and Adaptation in Sports Strategies' showed mid-level prevalence, suggesting that research topics are becoming increasingly specialized. Lower prevalence topics included 'Sensor-based IoT and Biomechanical Training' and 'Medical Interventions and Exercise for Health Enhancement', showing attempts to expand sports research into technological areas or other academic fields. Second, an analysis of the changes in topic prevalence over the years indicated that most topics maintained relatively high prevalence, suggesting that sports AI research is in its early stages. Third, the analysis of the semantic coherence and exclusivity of the topics revealed that 'Sensor-based IoT and Biomechanical Training' had the highest coherence, while 'Digital Economy and Industry Evolution in Sports' showed the highest exclusivity. This indicates that these topics are distinctly differentiated from other topics. This study provides a comprehensive analysis of research trends in the application of AI in the field of sports and contributes academically by analyzing unstructured data using the STM technique.

키워드