Exploratory Project: Connolly Linked Open Data

This past October, we launched The Séamus Connolly Collection of Irish Music, a digital collection that features over 330 traditional tunes and songs collected by master fiddle player Séamus Connolly, Sullivan Artist in Residence in Irish Music at Boston College (2004 to 2015). In addition to the wonderful music, this collection contains rich metadata, which gives us the opportunity to identify and represent relationships between the 130 or so musicians using linked open data (LOD) vocabularies and ontologies.

LOD will give people access to a wider network of information about the musicians in The Séamus Connolly Collection of Irish Music that may be available elsewhere on the semantic web through services such as DBpedia, MusicBrainzVIAF, and elsewhere. Presenting the relationships in a dynamic network graph visualization will provide a unique way to see, explore, and access information about the musicians and relationships between musicians within this collection. The current project team is made up of Anna Kijas (Digital Scholarship), Ben Florin (Systems), Nancy Adams (eScholarship/Digital Publishing), and Kelly Webster and Meg Critch (Metadata).

One of the first steps we tackled was to identify ontologies that describe the relationships between musicians based on the stories and metadata in the collection.

Ontologies:
1. Vocab.org
Relationship type is identified using the vocab.org ontology. For example, if someone collaborates with another musician on a tune in the collection, we would identify this as rel:collaborates_with

2. The Music Ontology
To identify the composition (mo:Performance_of) on which the musicians collaborated, we are using The Music Ontology. This ontology is also being used to identify the specific instruments (mo:instrument) used in the performance by a musician.

3. FOAF
Each musician’s name is identified with foaf:name and whether they are a composer or performer using foaf:agent. Once we identified these ontologies, Nancy and Anna broke out the metadata and began identifying the relationships for each musician. In addition to the ontologies, we added columns where we could add URIs for each musician from one or multiple commonly used authority services or structured data communities: VIAF, MusicBrainz, and DBpedia.

Record for release in MusicBrainzNot every musician had a URI associated with them; therefore, we decided to create records for them in MusicBrainz (a community-maintained open source encyclopedia of music information). Kelly and Meg experimented with creating records in MusicBrainz. They started by creating an artist record and linking it to an individual recording in the Connolly Collection. They also uploaded the entire track listing of The Séamus Connolly Collection of Irish Music into MusicBrainz and are currently creating links between the artists and the recordings within MusicBrainz. Once the artist records were created in MusicBrainz Nancy added URIs to our data.

Audio track using SoundCloud

In order to make the information about each musician human and machine readable we will generate RDF pages. We currently have an example RDF page for one of the musicians – Gabriel Donahue – which illustrates the ontologies, vocabularies, and URIs referenced above in addition with schema-org vocabulary that will identify each audio track that can be accessed via SoundCloud.

Once the data review is complete we will call on Ben’s expertise to auto-generate the RDF pages for each musician, as well as to create N-Triples and network graphs visualizing the relationships. Stay tuned for more updates!