ETD

• ai
• artificial intelligence
• community detection
• multilayer network
• multiplex network
• music discovery
• network science
• social network analysis

This thesis applies multiple network analysis to last.fm data (120,000 users and 4,028 artists) to detect music communities. We construct two-layer networks capturing behavioral similarity (shared listeners) and thematic similarity (lyrical content via textual forma mentis networks), comparing flattening, layer-by-layer (M-EMCD) and multilayer (Glouvain) community detection algorithms. The work reveals how communities emerge from the interaction between listening patterns and thematic preferences.