Sustainable Dissemination of Digital Music Artworks on TikTok: A Social Media Analysis

Abstract

This study examines sustainable dissemination of digital music on TikTok by linking musical-aspect discourse and network positioning to engagement and recommendation performance, using 1200 TikTok songs/videos, each represented by 52 platform and interaction features. Using a hybrid CF + NCF recommender with time-series forecasting, the predictive pipeline achieved F1 = 0.7879, AUC = 0.8373, and root mean squared error (RMSE) = 0.2143, while the ensemble forecaster yielded the lowest MAE = 0.2338 and RMSE = 0.3287 relative to ARIMA and Prophet. Content-side evidence showed that lyric-related sentiment was most prominent (84.91% positive mentions), exceeding production (52.78%) and melody, and matching results similarly prioritized lyrics over other musical aspects (exact: 30 vs. 19 vs. 17 and fuzzy: 53 vs. 39 vs. 29). Network analysis identified 1177 users and 164,499 ties partitioned into four communities, with the most bridging actor in Community 0 reaching betweenness centrality of 0.021, indicating concentrated brokerage in dissemination pathways. Together, these results suggest that lyric-centered engagement and community brokerage co-occur with stronger predictive and recommendation performance, providing an empirical basis for designing more durable TikTok dissemination strategies for digital music.