Re: [OEP] New draft of the note on n-ary relations

Dan,

Thanks for the note -- this is certainly a nice use case. I agree with  
you that dateranges is a common case of when you would need an n-ary  
relation. We have explicitly stayed away from it though, since  
representing time always tends to raise heated discussions and we felt  
it was not critical to the note. That said, there is no reason why you  
wouldn't use the approach that we outlined in the note (most probably,  
pattern 1) to represent the relations you are talking about.

Natasha

On Jan 26, 2005, at 1:53 PM, Dan Brickley wrote:

> Good to see this progressing (unlike my various TODOs, embarrasingly).
>
> I think I have an interesting use-case for you.
>
> The MusicBrainz project has a lot of metadata about... music.
>
> Artists, Tracks/CDs, Albums etc. Leigh Dodds has been working on
> a schema to go on the MB site; currently it serves RDF descriptions
> of albums etc but there's nothing at the namespace. Recently the MB
> team have come up with something that builds nicely on their core
> dataset, an "Advanced Relationships" system that allows MB contributors
> to catalogue a broader range of relationships amongst the entities
> MB knows about.
> http://blog.musicbrainz.org/archives/2004/12/advanced_relati.html
>
> example:
> http://test.musicbrainz.org/mm-2.1/artistrel/4d5447d7-c61c-4120-ba1b- 
> d7f471d385b9
>
> This is imho interesting for n-ary, because the system internally
> keeps track of more than simple binary relationships. Some  
> relationships
> are qualified with date ranges. For example, two artists were married
> between certain dates, or one artists played in another group between
> certain dates.
>
> I've not studied the WD and editors WD as carefully as I should, but
> from my experience this scenario (wanting to have an idiom for  
> recording
> '"x some-rel y" held between some-dates') is quite a common one. People
> ask me for similar on the FOAF list periodically, for example.
>
> Do you have any advice for how MusicBrainz might proceed?
>
> This btw is likely to be a dynamically growing namespace/dataset,
> as new relationships are added to the MusicBrainz database. I don't
> think we need to worry overly about formally capturing the intent
> w.r.t. temporal qualifications; rather, we just want a nice way
> of writing that info down, so that applications can get to it
> again.
>
> Thanks for any thoughts,
>
> Dan
>

Received on Monday, 31 January 2005 23:12:12 UTC