Tag Archives: visualization

We have made some strides since the last post where I described our initial work to represent relationships between musicians in The Séamus Connolly Collection of Irish Music (SCCIM) as linked data and a network graph. Recently, Kelly and Meg completed their work in creating links between the artists and recordings in MusicBrainz and this has resulted in a rich set of metadata. Each artist and instrument is now connected to each relevant recording.

Fig. 1 Connolly data in MusicBrainz

This results in linked records with aggregated data that shows all of the recordings a specific musician is connected to and the instrument they performed on per recording.  

Fig. 2 MusicBrainz record for Tina Lech

Another goal of this project is to create a network graph of the musicians in the SCCIM. To visualize the relationships between each musician, I am using Gephi, an open source network visualization tool, to generate the network graphs and render them with the Sigma.js library.

A force directed layout (ForceAtlas2) was applied to render this network graph. This layout is useful for smaller graphs, such as the SCCIM network, which has 158 nodes and 224 edges. Within the network graph, the musicians are represented as nodes (dots) and relationships are edges (lines). A musician is connected to other musicians by an edge if they performed together on one or multiple tunes. The network also shows single nodes, which represent a musician who composed and/or performed the tune as a soloist. All of the data was drawn directly from the metadata in the SCCIM.

Fig. 3 Snapshot of edge data

Fig. 4 Group legend

A color is assigned to each node in order to represent groupings by role. Group 1 includes musicians who are only performers; Group 2 includes both performers/composers; and Group 3 includes only composers. Relationships are defined as a collaboration between musicians on a tune within the SCCIM collection. 

Fig. 5 Sidebar view

When a node is selected, a sidebar opens up on the right with data about the musician and relationships within this network. Their name and role is listed, as is the degree (number of edges), which tells you how many collaborations this specific musician has within the context of the SCCIM. The musicians they collaborated with appear as links under “Connections.” When one of these names is clicked, the graph adjusts to show their connections. If Alice Bérubé is selected, for example, you’ll see that she collaborated with four different musicians, three of these are performers (Jeannine Webb, Pete Sutherland, and Ken Perlman), and one is a composer/performer (Seamus Connolly). Bérubé composed the tune “Don’t Get Me Anything” and also played the fiddle with Webb (fiddle), Perlman (banjo), Sutherland (piano), and Connolly (fiddle).

Fig. 6 Alice Bérubé’s connections

We focused solely on collaborations, because it was not possible to identify other types of relationships, such as who may have influenced who, who someone studied with, or band membership, because this data was not part of the original collection. In some cases, the stories written by Connolly shed some light on these additional relationships, for example, it might be mentioned that one of the musicians was a student of a certain individual or that they were influenced by a specific musician. This information might be useful to scholars and musicians interested in learning more about the way that musical traditions are transmitted, taught, and shared in the traditional Irish music community.

Visualizing the musicians who contributed compositions, performances, or both as a network graph provides a bird’s eye view at the distribution of roles. Using the “Group Selector,” you can easily see that there are 91 performers, 43 musicians who are both composers/performers, and 24 composers in this network. You can view the makeup of each group by selecting one specific group. These views depict the level of collaboration by role.

A longer term goal of this project is to not only show a general list of connections for each node, but to provide further context through the use of RDF and LOD so that users will be able to see how many times a musician interacted with another musician, on which tune(s), and which instrument they played, or whether they composed the tune that was performed. This data, which will live in RDF pages, such as this example, would be dynamically generated and accessed when someone selects a musician (node). This project is ongoing and additional updates will be shared along the way. You can view the version of the interactive network graph as described in this post online.