Hi, I’m Jakub Vrána, a developer at ČSFD, the Czech film database. Besides films, series and their creators, ČSFD includes a curated database of film soundtracks: albums entered by our editors with full tracklists, track durations and links to the films they belong to. I’d like to run a bot, vrana_bot, that adds IMDb links to soundtrack release groups.
What it does
It adds an IMDb relationship to about 7,100 release groups that don’t have one yet. Each one is paired with a soundtrack in ČSFD:
- the release name matches the soundtrack name (ignoring suffixes like “(Original Motion Picture Soundtrack)”) and the track count is the same,
- the tracklist matches: at least 80 % of titles in order (5,900 of them all titles) and, where both sides have durations (6,700 of them), within 3 seconds per track on average, mostly under 1 second,
- the ČSFD soundtrack belongs to exactly one film, and Wikidata gives the same IMDb ID for that ČSFD film, so the ID is current and not a merged duplicate,
- release groups that already have an IMDb link are skipped, and so are release groups matching soundtracks of different films, seasons and episodes, and short tracklists (5 tracks or fewer) without matching titles and durations.
About 6,600 of the release groups already have the Soundtrack secondary type; the rest are mostly untyped TV series soundtracks.
Edit note
Every edit explains the evidence, e.g.:
IMDb link from ČSFD, the Czech film database: https://www.csfd.cz/film/68969/soundtracky/
The tracklist of https://musicbrainz.org/release/29ba51ce-475b-4cab-aefe-fb33dd0b6a07 matches the ČSFD soundtrack “Wonder Woman” (130 tracks, 100% of titles, durations differ by 0.0 s per track on average), which belongs to that film only, and Wikidata gives the same IMDb ID for the ČSFD film.
Sample
I have made 11 test edits: https://musicbrainz.org/user/vrana_bot/edits. Sorry, I only read afterwards that the bot type has to be set first; I’ve stopped until it is. Could someone from the team set vrana_bot’s user type to Bot?
Pace
At most 1,000 edits per day and never more than 2,000 open edits, so about a week or two in total. The edits go strongest evidence first. I’ll read and answer edit notes on the bot’s edits.
Code
The source is at vrana_bot: adds IMDb links to MusicBrainz release groups whose tracklist matches a ČSFD soundtrack · GitHub, also linked from the bot’s user page.
Any feedback on the matching or the edit note is welcome before I continue.
