LinkedBrainz

For anyone following up on the LinkedBrainz work, the following links should guide through the status and give an introduction to the work.

https://github.com/BarryNorton/D2R-LinkedBrainz-Fork
project repository that includes
- the D2RQ mapping file [1]
- the libraries his D2RS instance depends on [2]
- a wiki with LinkedBrainz dataset related information [3]
- an issues list with open issues that are discovered during the development of the mapping [4]

https://github.com/zazi/linkedbrainz-d2rs-translators
project repository that includes
- the LinkedBrainz related D2RQ Translator classes that are necessary for the mapping [5]
- an issues list with open issues that are not resolved yet [6]

fluidsOps AG have taken the OWLIM database, loaded with the LinkedBrainz dump, and created a public SPARQL endpoint. This is used, together with their Information Workbench product to provide a front-end user interface for LinkedBrainz.

Discussion at JISCexpo Programme Meeting

Present: Barry Norton, Cedric Mesnage, Joe Lambert, David Bretherton, Simon Dixon

Identifiers for musical works
- Too much for one project -> crowd-sourcing
- Build on MusicBrainz infrastructure?
- Use MusicNet's alignment tool to clean up MusicBrainz (MB)
- or do the MB editors do a good enough job?
- do we need to involve musicologists in MB as super-editors
- MB only covers recorded music
- Involve publishers for scores without recordings?
- BBC Radio 3 data (Nick Humphrey?)

Bootstrapping issue to attract people to site (cheese)
- does MB already provide the right interface? could write a plugin for Picard
- is there any conflict between MB's sustainability and our specialist needs
- discography = cheese

Title of Primary Project Output: LinkedBrainz Live!

The type of user this product is for: LinkedBrainz Live is for application developers and institutions who require or use information about recorded music. Especially if thes require to interlink datasets or benefit from inference via automated reasoning.

With the LinkedBrainz D2R fork running under the latest revision of mappings we have produced an dump of the current RDF translation of the MusicBrainz NGS dataset. This is now 23.8GB of NTriples, ~180M assertions, which are available as a 2.7GB gzip compressed download.

In order to load this dump it is necessary to chunk it into manageable pieces. For instance, to load into OWLIM execute the following:
$split -l 1000000 musicbrainz_ngs_dump.rdf.ttl.gz $owlim/preload