Suggestion for widening MB Data Schema to address many artist issues.

[Vivek Kushwaha]

Hi,
This being my first post on MB mailing list, please tolerate all my deviations for observed norms, if any. After my post on forum, someone suggested my to make a post here. Goals being discussion and possible implementation, following is my post.

---

With the current MB schema for artists there are several issues which go unaddressed. I would like to throw a light some of the problems and suggest a better scheme.

Case of Mama Cass: There are (solo) artists who have released albums under different stage name over the course of their career. E.g. Mama Cass (Real Name: Ellen Naomi Cohen) used Cass Elliot in her later albums. We end up with three different entries and albums under two of them. Someone often ends up adding album at the wrong place because people she remained Mama Cass for most even when she hated being called that.

Case of Siouxsie Sioux: Born Susan Janet Ballion, for most of the 80s, 90s half of 2000s she called herself Siouxsie Sioux. Beginning from "Siouxsie and the Banshees" to "Siouxsie & Morrisey" collaboration; we all would say "Siouxsie Sioux" was a member of .... When going solo now she opted for "Siouxsie." Again we ended up with a lot names and some people will be saying we can simply rename entry for "Siouxsie Sioux" to "Siouxsie" 'coz there isn't going to be any release cataloged under "Siouxsie Sioux."

I have looked at the NGS plan and they are already looking for method to make collaborations easier by use of artist_credit,which I think won't help us much.

We should look back at our data type and distinguish them based on the type of entities they are and then try to determine what exactly is the string which will go into artist in our music tracks.

Fundamentally we have three types of entities.

1. A Person: As in real life, with a date of birth, place of birth, etc.

2. A Stage Persona: acquired by (one and exactly one) person, with a name which may be different from his real name.

3. A band: which is an ensemble of one or more stage persona. This is what we want to be there in our tags. They have a date of formation, date of disbandment etc. This is the entity which releases albums.

we can go for relationships as follows:
person one-to-many stage persona
stagePersona many-to-many band

we can provide new features in site browsing where one can access a page like h*tp://musicbrainz.org/person/<personID>.html and can access to all his work cronologically in tables as
<year> | <title (RG)> | <band> | <stagePersona> | <type-of-release>

There are endless new things that can be worked up. All I wanted to say is that this kind of structure will provide more robust framework to work with and will be able to handle even more complicated artist/band relationships that we do now; and with lesser redundancy.

I'm not a SQL guru or anything, so whoever is working with MB database will have to chalk out the implementation, joins, number of tables etc.

This idea just sprouted out of a problem I had with edit on "Siouxsie Sioux." I'm sure others have faced other limitations in current system and discussed it on edit notes with or without result. I've also forgot a few things that were in my mind before about this artist scheme.

I guess it will get a wider audience in forum and might help in improvement of MB rather than staying in my mind for some more time and then be lost for ever.

Regards,
Vivek
@MB site/forum: yoursvivek

--
0001 00 0001 0 101   101 001 000 0000 011 01 0000 01

_______________________________________________
MusicBrainz-devel mailing list
MusicBrainz-devel@...
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-devel

[Brian Schweitzer]

It's late, and I've been up for far too long, so only a short response.  :D

I'd suggest that the breakdown you present is still too simplified.  Take your #2, as example;  you define it as "acquired by (one and exactly one) person, with a name which may be different from his real name."  However, there are also the cases where you have a stage persona which more than one person may voice - this is quite common in the fictional artists, where you may have a cartoon character (or muppet) who has had multiple people provide the voice over time, or where one person may provide the speaking voice, while another provides the singing voice.

Or, in respect to bands, one issue I've been running into frequently lately, which is sub-groups; the closest to a correct description I can figure is to misuse (or is it a misuse?) the "member of" AR, such that we describe group X with subgroup Y as "(group) artist Y is/was a member of (group) artist X".  Subgroups too would seem to fall outside of the breakdown you provided.

As for the problem you suggest initially, if I understand correctly, artist_credit is being introduced to solve exactly that problem - you have an artist "Siouxsie Sioux", who has an AC "Siouxsie"; on track X of release A, the credit on the liner is to "Siouxsie Sioux", but on track Y of release B the credit is given to "Siouxsie".  As for collaborations, that's an entirely different beast; "Siouxsie & Foo Bar" no longer is a fake workaround artist, but is instead made up of five things: "Siouxsie Sioux" (the artist), "Siouxsie" (the AC used by that artist in this collab), "&" (the particular separator used in the collab), "Foo Bar" (the other artist), and "Foo Bar" (the AC used by the 2nd artist; in this case the AC being identical to the artist's 'true' name).

Brian

On Mon, Oct 26, 2009 at 8:03 PM, Vivek Kushwaha <yoursvivek@...> wrote:

Hi,
This being my first post on MB mailing list, please tolerate all my deviations for observed norms, if any. After my post on forum, someone suggested my to make a post here. Goals being discussion and possible implementation, following is my post.

---

With the current MB schema for artists there are several issues which go unaddressed. I would like to throw a light some of the problems and suggest a better scheme.

Case of Mama Cass: There are (solo) artists who have released albums under different stage name over the course of their career. E.g. Mama Cass (Real Name: Ellen Naomi Cohen) used Cass Elliot in her later albums. We end up with three different entries and albums under two of them. Someone often ends up adding album at the wrong place because people she remained Mama Cass for most even when she hated being called that.

Case of Siouxsie Sioux: Born Susan Janet Ballion, for most of the 80s, 90s half of 2000s she called herself Siouxsie Sioux. Beginning from "Siouxsie and the Banshees" to "Siouxsie & Morrisey" collaboration; we all would say "Siouxsie Sioux" was a member of .... When going solo now she opted for "Siouxsie." Again we ended up with a lot names and some people will be saying we can simply rename entry for "Siouxsie Sioux" to "Siouxsie" 'coz there isn't going to be any release cataloged under "Siouxsie Sioux."

I have looked at the NGS plan and they are already looking for method to make collaborations easier by use of artist_credit,which I think won't help us much.

We should look back at our data type and distinguish them based on the type of entities they are and then try to determine what exactly is the string which will go into artist in our music tracks.

Fundamentally we have three types of entities.

1. A Person: As in real life, with a date of birth, place of birth, etc.

2. A Stage Persona: acquired by (one and exactly one) person, with a name which may be different from his real name.

3. A band: which is an ensemble of one or more stage persona. This is what we want to be there in our tags. They have a date of formation, date of disbandment etc. This is the entity which releases albums.

we can go for relationships as follows:
person one-to-many stage persona
stagePersona many-to-many band

we can provide new features in site browsing where one can access a page like h*tp://musicbrainz.org/person/<personID>.html and can access to all his work cronologically in tables as
<year> | <title (RG)> | <band> | <stagePersona> | <type-of-release>

There are endless new things that can be worked up. All I wanted to say is that this kind of structure will provide more robust framework to work with and will be able to handle even more complicated artist/band relationships that we do now; and with lesser redundancy.

I'm not a SQL guru or anything, so whoever is working with MB database will have to chalk out the implementation, joins, number of tables etc.

This idea just sprouted out of a problem I had with edit on "Siouxsie Sioux." I'm sure others have faced other limitations in current system and discussed it on edit notes with or without result. I've also forgot a few things that were in my mind before about this artist scheme.

I guess it will get a wider audience in forum and might help in improvement of MB rather than staying in my mind for some more time and then be lost for ever.

Regards,
Vivek
@MB site/forum: yoursvivek

--
0001 00 0001 0 101   101 001 000 0000 011 01 0000 01

_______________________________________________
MusicBrainz-devel mailing list
MusicBrainz-devel@...
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-devel

_______________________________________________
MusicBrainz-devel mailing list
MusicBrainz-devel@...
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-devel

[Robert Kaye]

On Oct 26, 2009, at 5:03 PM, Vivek Kushwaha wrote:

> Case of Mama Cass: There are (solo) artists who have released albums  
> under different stage name over the course of their career. E.g.  
> Mama Cass (Real Name: Ellen Naomi Cohen) used Cass Elliot in her  
> later albums. We end up with three different entries and albums  
> under two of them. Someone often ends up adding album at the wrong  
> place because people she remained Mama Cass for most even when she  
> hated being called that.
>
> Case of Siouxsie Sioux: Born Susan Janet Ballion, for most of the  
> 80s, 90s half of 2000s she called herself Siouxsie Sioux. Beginning  
> from "Siouxsie and the Banshees" to "Siouxsie & Morrisey"  
> collaboration; we all would say "Siouxsie Sioux" was a member  
> of .... When going solo now she opted for "Siouxsie." Again we ended  
> up with a lot names and some people will be saying we can simply  
> rename entry for "Siouxsie Sioux" to "Siouxsie" 'coz there isn't  
> going to be any release cataloged under "Siouxsie Sioux."
>
> I have looked at the NGS plan and they are already looking for  
> method to make collaborations easier by use of artist_credit,which I  
> think won't help us much.

I disagree -- for each of these releases we can create an  
artist_credit that is correct for that particular release. These will  
in turn then link back to a canonical artist that will have the  
artists name as they want to have the artist credited. I think NGS  
will handle all of these cases just fine

And logistically speaking, we need to finish NGS and get it out the  
door -- which means that we can't go back to re-think anything right  
now. I would suggest that you sit tight for the time being and when we  
make a push for beta 1 on NGS that you go and see if NGS really can/
can't handle these cases adequately.

If not, we can discuss this for our post NGS roadmap.

--

--ruaok        The answer to whether or not something is a good idea
or not, should not be taken as an indication of whether I want to do it.

Robert Kaye     --     rob@...     --    http://mayhem-chaos.net









_______________________________________________
MusicBrainz-devel mailing list
MusicBrainz-devel@...
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-devel

[Chad Wilson-2]

-------- Original-Nachricht --------
> Datum: Tue, 3 Nov 2009 13:57:24 -0800
> Von: Robert Kaye <rob@...>

> I disagree -- for each of these releases we can create an  
> artist_credit that is correct for that particular release. These will  
> in turn then link back to a canonical artist that will have the  
> artists name as they want to have the artist credited. I think NGS  
> will handle all of these cases just fine

FWIW, this is the same as what I said on Vivek's earlier crosspost to the forums: http://forums.musicbrainz.org/viewtopic.php?id=1940 and also mentioned about what would be cool in how taggers could use these credits vs. canonical names in future.

Cheers
Chad
--
GRATIS für alle GMX-Mitglieder: Die maxdome Movie-FLAT!
Jetzt freischalten unter http://portal.gmx.net/de/go/maxdome01

_______________________________________________
MusicBrainz-devel mailing list
MusicBrainz-devel@...
http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-devel