This year’s meeting of the Music Library Association was held in Cincinnati, where, during breaks and receptions, we enjoyed 1920s tunes performed by members of the Cincinnati Opera, and by MLA’s own big band, in the Netherland Plaza Hotel’s beautifully-restored 1930 Art Deco ballroom.


It has long been recognized that America’s conservatories and orchestras remain overwhelmingly white (less than 5% of students in music schools are non-Asian minorities). While administrators of these institutions are currently struggling to rectify the situation, libraries (it was noted at the MLA meeting) have a chance to be an exemplar. In a joint project with ARL called the Diversity & Inclusion Initiative, MLA has supported internships and fellowships for MLIS students with music backgrounds to work in music libraries. The diversity aimed for includes not just race/ethnicity, but also gender, marital status, disabilities, etc. In the opening plenary session, we heard from some of the former fellows. Benefits that were particularly appreciated included the visibility and recognition acquired while a student, which subsequently opened doors to professional opportunities; peer mentors (previous fellows) who provided ongoing support with entry into the profession, and after; and help with the hidden costs of college (additional fees, textbooks, etc.) for which first-generation students are often unprepared. Difficulties encountered included locating sources of help – one fellow reported “cold calling” random MLA members before discovering the DII program. This prompted a discussion, during the Q&A, on how the program could be better publicized.

On a similar outreach note, MLA (whose membership encompasses North America – U.S. and Canada) plans to invite Latin American colleagues to next year’s meeting in Orlando, billing it a Pan-American conference.


MLA’s initiatives in this field:

  • Two new thesauri have been published in the past year — for medium-of-performance terms (LCMGT), and for music genre/form terms (LCGFT) – along with best-practices documents for both.
  • Involvement in LD4L (Linked Data for Libraries), a collaborative project of Cornell, Harvard, and Stanford.
  • The NACO Music Project, working on authority data.
  • A Bibframe Task Force, which is undertaking various projects to enhance the new encoding schema to meet music users’ needs.

We heard about other projects that member libraries have done to enhance discoverability of special collections:

The Linked Jazz Project, best known for its visualizations, is based on data extracted from oral-history transcripts in numerous jazz archives. The data is then converted to RDF triples reflecting relationships between jazz artists (x talks about y; y knows of x). The data is enhanced via crowdsourcing. The developers hope others will use the LJ data to build additional linked-data sets: mashing LJ data with performances at Carnegie Hall is one such project; another is unearthing female jazz artists (neglected in traditional jazz histories) by enriching LJ data with other sources such as DBpedia, MusicBrainz, and VIAF (the international authority file).

Colleagues at Michigan State used Discogs (a crowdsourced, expert-community-reviewed database of metadata on pop music recordings) to process a gift collection of 1200 LPs of Romani music, which also included pop music containing Gypsy stereotypes. They hope to use this collection as a pilot to develop a process for a much larger corporate gift of 800,000 pop recordings and videos. They were able to extract data directly from the Discogs website using Discogs’ API (which outputs in JSON – they used Python to convert the JSON to XML and then MARCXML). Cataloging challenges included: dealing with usage differences between Discogs’ “release” and RDA’s “manifestation”; similarly, between Discogs’ “roles” for artists and RDA’s “relationship designators”; and mapping Discogs’ genres and subgenres to LC’s genre/form terms and medium-of-performance terms, supplementing with LC subject headings as needed. Discogs’ strengths: expertise in languages (from its international contributor community) and in obsolete formats; and the ability to link to the Discogs entry from the library catalog. Our presenters plan to propose to the Discogs community indexing the UPC (universal product code, the barcodes on CDs); a similar resource, MusicBrainz, does this.

A third project, at Cornell, was ultimately unsuccessful, but also illustrates the variety of data resources and tools that people are trying to link up. For a collection of hip-hop flyers, they constructed RDF triples using data from MusicBrainz, ArtStor, and Cornell’s existing metadata on the related events etc. They chose Bibframe for their encoding schema, and compiled an ontology from Getty’s AAT vocabulary, various music and event ontologies, and Reconciliation of names from all these sources was done using the open-source analytics tool OpenRefine. The problems developed as they came to feel that Bibframe did not meet their test for describing flyers; they decided to abandon it in favor of LD4L. Reconciliation of names also proved more problematic than expected.


In a session on music discovery requirements, colleagues noted two things that current ILSs and discovery layers are not good at: showing hierarchies (for instance, making available additional search terms in thesauri, ontologies, etc.); and mapping multiple physical formats to one title (for multi-media items, such as a book issued with a disc, or a score with a recording, or a CD with a DVD – in most interfaces, the content of the second piece will not be retrieved under a format-facet search).

A presenter from Stanford proposed facet displays that include drop-down menus showing a relevant thesaurus, allowing users to further narrow to a subgenre, for instance. For music, the newly-developed medium-of-performance thesaurus, if displayed with multiple search instances, could enable musicians to enter all the instruments in their ensemble, and retrieve music for that specific combination of instruments. Also discussed were domain-specific search interfaces, such as the ones done by UVA for music and videos. Needless to say, there are potential applications for other disciplines.

Colleagues at East Carolina have made use of Blacklight to map multiple physical formats to the same title.