DETECTING COLLABORATION PROFILES IN SUCCESS-BASEd MUSIC GENRE NETWORKS Gabriel P. Oliveira, Mariana O. Silva, Danilo B. Seufitelli, Anisio Lacerda, Mirella M. Moro PROFILES Solid, Regular, Bridge and Emerging highlightshighlights COLLABORATION in general, genre collaborations are LOCAL MATTERS increasing local genres play a key role on determining hit songs introductionintroduction collaborations are more popular than ever 1 1 1 1 1 1 1 1 2 1 1 1 2 2 2 2 2 95 96 96 97 97 97 98 98 01 99 99 99 00 00 01 01 01 8 2 6 0 4 8 2 6 4 0 4 8 2 6 0 4 8 objectiveobjective We aim to unveil the dynamics of cross-genre connections and collaboration profiles in success-based networks Does the regional aspect impact on popular genres Rq1Rq1 and their hit songs? How has genre collaboration evolved over the past Rq2Rq2 few years? Which are the potentially intrinsic factors and Rq3Rq3 indicators that influence the collaboration success? METHODOLOGyMETHODOLOGy build MODEL DETECT Proper dataset Success-based Collaboration Genre Network Profiles Music GENRE dataset (mgd) Spotify API Data Weekly Song Charts Global and Regional songs, artists, genres from 2017 to 2019 top music markets mgd enhanced genre collaboration data 1,330 Charts 13,380 Hit Songs 3,612 Artists 896 Genres Does the regional aspect impact on popular genres and Rq1 their hit songs? most popular music genres in each considered market (2017 - 2019) Global Australia Brazil Canada GermLany France UK Japan USA GENRE COLLABORATION NETWORK Hit Songs nodes are divided into three sets: artists genres, artists, and hit songs i.e., the minimum elements to evaluate genres success Tripartite Graph 1 collaborative hit songs are sung by two or more artists, 1 1 regardless of their 1 participation (e.g., a typical feat. or a duet) Artist Network 2 Intermediate Step 1 one-mode network in which nodes are 1 2 exclusively genres, where edge weight is the 2 number of hit songs by 1 artists from both genres Genre Network 3 self-loops are allowed and represent intra-genre collaborations Rq2 How has genre collaboration evolved over the past few years? Network characterization 709 564 583 257 247 237 72 79 89 16 15 16 2017 20L18 2019 GROUPGROUP 11 9.21% 12.59% 4.72% # genres avg. degree avg. clustering coefficient GGRROOUUPP 22 3.17% 22.29% 20.15% # genres avg. degree avg. clustering coefficient GGRROOUUPP 33 6.76% 3.64% 2.12% # genres avg. degree avg. clustering coefficient GENRE collaborations profiles GENRE colExploratorylaborat factor analysis Preferential Attatchment F1 Attractiveness Common Neighbors Resource Allocation F2 Affinity Weight Neighborhood Overlap F3 Influence Edge Betweenness cluster analysis dbscan Clustering result USUSAA nenetworktwork (2019)(2019) Cluster 0 Cluster 1 Cluster 2 Cluster 3 7.5 rock - new wave hip hop - rap other - classical 5.0 rock - punk rap - pop rap rap - trap pop - folk rock rap - rap rock - experimental pop - lounge 2.5 pop - pop lounge - swing trap - pop pop - rock dance pop - rock Dim2 (17.7%) pop - other pop - r&b classic rock - experimental rap - r&b pop - reggaeton hip hop - dance pop dance pop - brazilian funk 0.0 pop - house indie rock - rap rock dance pop - dance pop latin - reggaeton new wave - punk latin - latin soul - soul k-pop - k-pop reggaeton - reggaeton moombahton - moombahton -8 -4 0 4 Dim1 (66.1%) collaboration profiling Collaboration profiles for all markets (2019) Preferential Common NeighborhooRdesourceEdge Resource Weight Attatchment Neighbors Overlap AllocatioBne tweenness Allocation Attractiveness Affinity Influence Which are the potentially intrinsic factors and indicators that Rq3 influence the collaboration success? Median Number of Streams 2017 20L18 2019 PROFILE 0 ssoolliidd 2017 20L18 2019 high Attractiveness and Affinity POP medium Influence successful genre collaborations HIP-HOP inter-genre collaborations PROFILE 1 RREGEGULULAAR 2017 20L18 2019 medium values POP successful genre collaborations SOUL inter-genre collaborations PROFILE 2 bribriddggee 2017 20L18 2019 high Influence low Attractiveness and Affinity GOSPEL less successful RAP ONLY INTER-GENRE COLLABORATIONS PROFILE 3 EMEMERGERGIINGNG 2017 20L18 2019 medium Attractiveness and Affinity k-pop low Influence k-pop less successful intra-genre collaborations base material further r research tasks, e.g., prediction and recommendation conclusionconclusion music industry maximize expected success by artists properly investing in potential profit by identifying the collaborations most suitable partnerships contactcontact usus gabriel p. oliveira mariana o. silva [email protected] [email protected] OUR PROJECT Danilo b. Seufitelli anisio lacerda mirella m. moro [email protected] [email protected] [email protected].