Discovering Metadata Inconsistencies

From AcaWiki
Jump to: navigation, search

Citation: Bruno Angeles, Cory Mckay, Ichiro Fujinaga (2010) Discovering Metadata Inconsistencies. ISMIR (RSS)
Internet Archive Scholar (search for fulltext): Discovering Metadata Inconsistencies
Download: http://jmir.sourceforge.net/publications/ISMIR 2010 Discovering.pdf
Tagged: musicbrainz (RSS)

Summary

Authors compare metadata from MusicBrainz, a manually curated library "Codaich" and a reference library of unprocessed files. With the exception of classical artist names there is a higher matching percentage (across classical, jazz, popular and world genres, matching for artist, album, and title) between MusicBrainz and Codalich than between MusicBrainz and the unprocessed files, possibly indicating the value of curated metadata (MusicBrainz and Codalich both curated, by crowd and experts respectively).

Various technical challenges are noted, including some which could affect results (eg the reference library used ID3v1 tags, limited to 30 characters).


Retrieved from "https://acawiki.org/index.php?title=Discovering_Metadata_Inconsistencies&oldid=9995"