Behavioral and Acoustic Artist Graphs for Music Community Detection and Genre Inference

• Manuel Bezerra-Brandao UNIFESP
• Lilian Berton UNIFESP

DOI: https://doi.org/10.5753/eniac.2026.30956

Resumo

This work compares behavioral and acoustic artist graphs for community detection and missing-genre inference. A behavioral graph is constructed from repeated artist proximity in ten Last.fm listening histories, while an acoustic graph is constructed from aggregated AcousticBrainz low-level descriptors. Genre labels, tags, textual metadata, and high-level acoustic annotations are excluded from graph construction; MusicBrainz genres are used for evaluation and downstream inference. Louvain, Leiden, Node2Vec, Graph Autoencoders, and propagation controls are evaluated. The observed behavioral results show higher adjusted genre alignment and Macro-F1; however, differences in artist coverage, genre spaces, and class distributions prevent direct comparison between views. Acoustic similarity instead provides a complementary, coverage-limited view of artist relatedness. In a stratified manual audit, considering the top three behavioral Node2Vec suggestions provided at least one genre judged useful for 90.7% of reviewed artists. These findings support graph-based candidate generation for human-assisted metadata curation.

Referências

Bedi, P. and Sharma, C. (2016). Community detection in social networks. Wiley interdisciplinary reviews: Data mining and knowledge discovery, 6(3):115–135.

Blondel, V. D., Guillaume, J.-L., Lambiotte, R., and Lefebvre, E. (2008). Fast unfolding of communities in large networks. Journal of Statistical Mechanics: Theory and Experiment, 2008(10):P10008.

Correa, D. C. and Levada, A. L. M. (2011). Finding Community Structure in Music Genres Networks. Poster Session.

Dolgikh, D. and Jelínek, I. (2015). Graph-based music recommendation approach using social network analysis and community detection method. In Proceedings of the 16th International Conference on Computer Systems and Technologies, pages 221–227, Dublin Ireland. ACM.

Fortunato, S. (2010). Community detection in graphs. Physics Reports, 486(3-5):75–174.

Grover, A. and Leskovec, J. (2016). node2vec: Scalable Feature Learning for Networks. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 855–864, San Francisco California USA. ACM.

Haupt, J. (2009). Last.fm: People-Powered Online Radio. Music Reference Services Quarterly, 12(1-2):23–24.

Kipf, T. N. and Welling, M. (2016). Variational Graph Auto-Encoders. arXiv:1611.07308 [stat.ML].

Porter, A., Bogdanov, D., Kaye, R., Tsukanov, R., and Serra, X. (2015). AcousticBrainz: A community platform for gathering music information obtained from audio. In Proceedings of the 16th International Society for Music Information Retrieval Conference (ISMIR 2015), pages 786–792, Málaga, Spain.

Stutzbach, A. R. (2011). MusicBrainz (Review). Notes, 68(1):147–151.

Traag, V. A., Waltman, L., and Van Eck, N. J. (2019). From Louvain to Leiden: guaranteeing well-connected communities. Scientific Reports, 9(1).

Vlegels, J. and Lievens, J. (2017). Music classification, genres, and taste patterns: A ground-up network analysis on the clustering of artist preferences. Poetics, 60:76–89.

Zheng, M., Domanskyi, S., Piermarocchi, C., and Mias, G. I. (2021). Visibility graph based temporal community detection with applications in biological time series. Scientific reports, 11(1):5623.