P-003 · 2025 · Team project
Artist influence networks with Neo4j
Graph analysis of musical influence from 1960 to 2020: SecondHandSongs cover data in Neo4j, scored with PageRank, in-degree and betweenness, then clustered.
Three views of influence
Centrality profiles clustered with Graph Data Science
The question
Which artists shaped popular music most between 1960 and 2020? We answered with one measurable proxy: who gets covered by whom. Every cover recorded on SecondHandSongs becomes a directed edge from the covering artist to the original artist, weighted by how often it happens.
Approach
We loaded artists, songs and covers into Neo4j and projected an artist-to-artist influence graph. Graph Data Science queries then score each artist in three ways. In-degree counts how often an artist is covered, PageRank also rewards being covered by artists who are themselves influential, and betweenness finds artists who connect otherwise separate groups. K-means clustering on these scores groups artists with similar influence profiles, and Neo4j Bloom makes the groups explorable.
Results
The Beatles lead on every measure. In the published graph they receive 9,389 cover links, more than twice Duke Ellington’s 4,486. Bob Dylan, Elvis Presley and the Rolling Stones also rank near the top, and the clusters roughly follow genres such as rock, jazz and pop.
My role and limits
I worked on this as part of a team project, and the repository contains the graph data, the Cypher queries and the Bloom setup. Influence here means recorded covers, so the ranking describes this graph rather than musical importance in general. Artists whose covers are well documented on SecondHandSongs may rank higher than their wider influence would justify.