Sports management is a field of study that has grown significantly. However, little is known about its structure. Therefore, this article presents an unsupervised cluster learning approach based on a thematic analysis of abstracts. It applies to a large corpus of sports management documents to gain insight into the structure of the discipline. A total of 3,704 records were downloaded from the Scopus database. Data were processed using the BERTopic approach. The results indicate that sports management is a growing and highly collaborative field of study. The themes (or topics) were diversified (25 clusters). Due to their proximity, seven meta topics were also identified. While the interest in major sports events and the social impact of sports tourism has lessened in recent years, other fields of study, such as entrepreneurship, innovation and sports club management, have captured the interest of researchers. Keywords: Sports management, bibliometric analysis, topic modelling, BERTopic, UMAP, hDBSCAN.