Live Music Archive Linked Data: Mapping
The etree dataset contains mappings to other collections such as MusicBrainz and geonames. These mappings have been derived using the following methods.
Artist Mapping
http://etree.linkedmusic.org/vocab/simpleMusicBrainzMatch
Method used to generate mappings between artists in etree and MusicBrainz. For a given artist A, a mapping to MusicBrainz is generated as follows:
- Query MusicBrainz via the API for artists matching name A.
- If there is a single result with score 100, consider this as a candidate.
- Determine lexical distance between the candidate and A. For this purposes we used the ruby amatch gem and a Jaro metric. If the distance is within a particular threshold (0.9) for the current data set, then assert a similarity.
Location Mapping
http://etree.linkedmusic.org/vocab/simpleGeoAndLastfmMatch
Method used to generate mappings between venues in etree, locations in geonames and venues in last.fm. For a performance with venue V and coverage L, we use geonames and last.fm to pinpoint geolocations for performances as follows:
- Query last.fm for venues with the given name V. Returns a collection {V_i}
- Query geonames for populated places with the name L. Returns a collection {L_i}.
- For each pair (V_i,L_j), check
- whether the lat/long coordinates coincide within a given threshold (currently 10 km)
- whether the location associated with V_i is lexically similar to the name of L_i\footnote{using a Jaro matcher with a threshold of 0.8}.
- If both conditions are met, then the location L_i and the venue V_i are considered candidates for mapping. In this case, L_i is asserted as similar using the method http://etree.linkedmusic.org/vocab/simpleGeoAndLastfmMatch
http://etree.linkedmusic.org/vocab/simpleGeoMatch
Method used to generate mappings between venues in etree, locations in geonames and venues in last.fm. For a performance with venue V and coverage L, a http://etree.linkedmusic.org/vocab/simpleGeoMatch similarity is asserted if the process above yields no last.fm venue match.
http://etree.linkedmusic.org/vocab/simpleLastfmMatch
Method used to generate mappings between venues in etree and venues in last.fm. For a performance with venue V and coverage L, a http://etree.linkedmusic.org/vocab/simpleLastfmMatch similarity is asserted if the process above yields a last.fm match.