Semantic Communities and Boundary-Spanning Lyrics in K-pop: A Graph-Based Unsupervised Analysis
The study used multilingual song embeddings to model K-pop lyrics as a graph. By applying community detection, researchers found 21 operational semantic communities without genre labels. Boundary-spanning songs, identified by network centrality and diversity metrics, act as bridges between these themes. This framework reveals how lyrics organize into meaning structures and identifies songs that connect different thematic regions.