Our goal is to help one find unexpected connections in music making. To find that a Hard Rock drummer from a record you love has recorded with a Samba band, for example. This can assist a listener in exploring a collection, discover new music, or to dive in different aspects of an artist's work.

Our approach is to use data from Discogs, inspired by a hack from HAMR 2014a. We process their monthtly dump about album releases to create a network that connects artists who a same collaborator in their records. For example, if Ron Carter recorded with Tom Jobim and Joe Henderson (and he did), these two artists will be connected with an edge that contains this information.

Using this network, we look for community structure in it, which shows groups of artists who have worked frequently with the same collaborators. We use Gephi for the Modularity algorithm (aka GLay) to identify communities and for visualizing the network. Gephi also allows us to export a Sigma.js network visualized which is put online.

Inspecting the communities and links between artists in different communities then lead one to discover unexpected links. Looking for their impact is then the fun left as an exercise for the reader.

During HAMR 2014 we have produced the networks below, looking (1) at brazilian artists in Discogs as connected through any kind of common collaborators, and (2) jazz artists in Discogs as connected through bass players shared.

Online visualization of brazilian artists: http://nazareno.github.io/unexpected/ . For bassists, go to http://nazareno.github.io/unexpected/bassists. Some more pictures below.

All Brazilian artists:

Connections only considering bossa nova, pop/soul and samba jazz communities (as found through GLay):

Jazz connections through bassists: