Genre and Version
kosm...@gmail.com
1.) Was genre intentionally left out?
2.) I see the Other Versions Relationships but none of these seem to
address encoding or live versions. Perhaps there is already a way to
express these properties. It seems, though that other_release_of and
remaster_of don't really cover it, though.
I see three levels to a musical work:
The piece
The performance
The artifact
This structure provides a way to talk about radio edits, live
performances, different concerts and the media that they are captured
in. I haven't spent the time to write up much formally in this vein but
it seems like this might be helpful. It would separate releases on
different media (LP, cassette, CD, etc) as well as various digital
encodings (MP3 CBR @ 128kbps, OGG @ 160kbps, FLAC, etc).
Oh, one last thing regarding genre: take a look at the metadata that
allmusic.com tracks. Some more properties to consider:
For artists:
genre
styles
moods
instruments
For albums:
type (format)
release year/date (DC?)
label
label catalog number
genre
styles
moods
description (external ont?)
instruments
For tracks:
composer
duration
description (external ont?)
instruments
genre
style
mood
bpm?
time signature?
Just some thoughts,
David Sheets
Frederick Giasson
> 1.) Was genre intentionally left out?
Yeah well, the first version is mainly (only?) based on musicbrainz. Since
musicbrainz doens't take care about the genre, the Music Ontology do not neither
at the moment.
However, it is now the time to look at things like this since a first version
exist and that we can build-up something much better from that starting point.
This is for that exact reason that a google group has been opened with that new
ontology :)
So yeah, we should start to think about how to describe genres in that ontology.
However, I'll come back later on that subject.
> 2.) I see the Other Versions Relationships but none of these seem to
> address encoding or live versions. Perhaps there is already a way to
> express these properties. It seems, though that other_release_of and
> remaster_of don't really cover it, though.
Take a look at the class mo:Live [1]:
Live - A release (mo:Album) that was recorded live.
So, how could you describe the fact that an Album is in fact a Live Album? I
thought at that:
<mo:Album
rdf:about="http://mm.Music.org/album/65f4f0c5-ef9e-490c-aee3-909e7ae6b2ab">
<rdf:type rdf:resource="http://purl.org/ontology/mo/Live" />
</mo:Album>
So, from the Album instance, you use the rdf:type property to say that this
instance is not only a mo:Album but also a mo:Live (so a mo:LiveAlbum at the end
;) ).
So the question now is: is it the good way to do that? It was the way choose by
Leigh when he developed the Musicbrainz Metadata Ontology 3 years ago.
BTW, Alex suggested to change mo:Album for something like mo:Release, and I
think it is a really good idea, since a live show is not really an album,
specially if the show is put on youtube after the live performance :)
> I see three levels to a musical work:
> The piece
> The performance
> The artifact
>
> This structure provides a way to talk about radio edits, live
> performances, different concerts and the media that they are captured
> in. I haven't spent the time to write up much formally in this vein but
> it seems like this might be helpful. It would separate releases on
> different media (LP, cassette, CD, etc) as well as various digital
> encodings (MP3 CBR @ 128kbps, OGG @ 160kbps, FLAC, etc).
Read what Alex Passent said later in the mailling list. You are right, and I
will propose something later today.
> Oh, one last thing regarding genre: take a look at the metadata that
> allmusic.com tracks. Some more properties to consider:
>
> For artists:
> genre
Agree
> styles
what is the difference between genre and styles? (In french it is quite the same)
> moods
Do you have a list of moods in mind?
> instruments
Should we list all existing instruments from the guitar to the melotron?
> For albums:
> type (format)
Agree and Alex too.
> release year/date (DC?)
Yeah, I already use dcterms:created for that purpose (you can take a look at the
RDF examples).
> label
> label catalog number
> genre
Agree
> styles
> moods
Same questions that for Artist
> description (external ont?)
using dc:description no?
> instruments
Same as for artists.
> For tracks:
> composer
You can infer it via mo:Album
So, this is a tradeoff we have to take:
More triples in the triple store or biggest sparql requests that can take more
time to process?
> duration
mo:duration already exists (in milliseconds)
> description (external ont?)
dc:description too, no?
> instruments
> genre
> style
> mood
same.
> bpm?
Could be.
> time signature?
What you mean?
[1] http://pingthesemanticweb.com/ontology/mo/#term_Live
So, great! thanks for all these suggestions: it is a good starting point to
improve the ontology.
I will write another email today that take your suggestions and the ones of
other people. If people agree, I will change it accordingly.
Take care,
Fred
kidehen
Fred,
See: http://islab.dico.unimi.it/ontologies/mxonto-genre.owl :-)
It's certainly worth integrating into this effort.
Kingsley