[Playlist] Track ambiguation

[Paul B Lamere]

Chris:

The grabb.it site looks pretty interesting ... looking forward to seeing 
it go live. 

I believe there's a 'AlternativeVersion RelationShipClass' that does 
what you want. It's described here:

http://wiki.musicbrainz.org/AlternativeVersionRelationshipClass

I'm guessing though that you won't find it to be very universally 
applied (there are others here that know much more than me about 
this).   However, it may be that you can synthesize some of the behavior 
that you are looking for ... such as looking for tracks that have the 
same length ... or looking for tracks that have colliding PUIDs (but I'm 
sure you've thought about this sort of thing already).


Paul


Chris Anderson wrote:

> Hey all,
>
> I'm new to the list here - I've been using XSPF for some time now
> (originally for Music For Dozens, and now with my work on Grabb.it )
> and I've finally found this list. So far every post that has come down
> the pipe has spoken to exactly the sorts of things we've been
> struggling with as we model Grabb.it's metadata domain.
>
> The service we'd like to provide is something similar to to Paul's
> playlist resolver. Eg. We'd like to take a user supplied playlist and
> resolve it into one for which we can locate playable entities. If the
> user supplied metadata is high-quality enough, it is easy to find
> MBIDs for list items, which correspond to individual releases of
> particular tracks.
>
> What we'd like to do is go one step further. We assume our users don't
> care about the difference between all the different releases of a
> given track, as long as the track is basically the same - eg. Iggy
> Pop's "Lust for Life" in original release, or on any one of the many
> compilations it was later re-released on, should be ambiguated into
> one Track (or MetaTrack ...) so we have the highest chance of finding
> a playable entity related to the requesetd track.
>
> It's not as easy as matching only on Artist and Title, because we
> don't want to end up with say, the Unplugged version of the track, if
> we can help it.
>
> So with that background, my question is - Is there any provision in
> the MusicBrainz system to say that two Tracks *may as well be the
> same* (from the perspective of the casual listener)? If not, is this a
> good candidate for an AdvancedRelationship?
>
> Let me know if this is better taken to another list.
>
> Thanks for any feedback!
> Chris
>