Possible community cleanup targets

Hi to all classical editors!

@ListMyCDs.com reminded me that we had talked already ages ago about trying to get some community cleanup off the ground in order to improve our information on major composers.

I’ve cleaned up a lot of composers on my own, and I know I’m not the only one. By “clean up”, here, I mean linking all the recordings that can be identified to works, finding performers for as many recordings as possible and setting the rest to [unknown], in order to have a composer entry where the only thing under “Recordings” are recordings where the composer actually played the music. For example, see Arvo Pärt or [Olivier Messiaen] (Olivier Messiaen - Recordings - MusicBrainz).

That said, it’s almost impossible to get someone like Liszt or Bach to that level without several editors working on it together. And we want to get editors more involved, so this seems like a good opportunity to do so. We’d want as many experienced editors as possible contributing (even if it’s just one album per day or a few every week!) but even for new classical editors who are still learning, this is a good option: there’ll be a strong support community in the forums, so they’ll be able to ask as many questions as they need and have other editors help with any mistakes or problems.

I’d like to hear people’s ideas about who would be a good first choice for an attempt at community cleaning. Unless people really want to do this only if it’s for a big composer whose music they really like, I’d rather avoid the biggest names (Bach, Beethoven, Mozart, Liszt, Haydn, Vivaldi, Chopin) for the trial run, and look into a “second tier” of popularity - I’m thinking Prokofiev, Debussy, Mahler (if the different editions of works aren’t too big an issue) or the like. So: ideas? :slight_smile:

5 Likes

One thing I’d suggest: let’s avoid composers who mostly worked on operas for the first go. Operas come with their own set of issues, and I’d like to do a similar project for one opera (probably by Verdi or Mozart) at some point to try and figure out the issues and see what we can improve, both editing and guideline wise, for the opera use case.

3 Likes

Another suggestion: perhaps not starting with a hugely popular composer with a huge output. A ‘smaller’ composer might be more manageable and we might even be able to ‘finish’ one (that is, cleaning up everything we already have in MusicBrainz, adding all releases with the composer’s work will be impossible).

4 Likes

I don’t have any specific composer in mind, but I was thinking that people who know how to edit classical music could maybe try to clean up some data that have been added with old guidelines (searching those having “feat. conductor” in release groups, releases, tracklists and recordings) and then cleaning up the whole release, adding proper relationships and so on. However searching this term yields so many results…

15,591 release groups contain "feat. conductor"
18,904 releases contain "feat. conductor"
488,136 recordings contain “feat. conductor”

3 Likes

The numbers aren’t quite this high – I think there’s about 2000 or so release groups with a conductor in the title. I try and fix some these when I find the time, even if you just fix the artist credit this is still a great help, as the release groups will then get listed in their performers’ discographies.

As for the composer cleanup, :+1: for Debussy.

2 Likes

Great idea!
I started doing statistics a few months ago to know how much work was left on the main composers, I should try to make some graphs and post them here

A few remarks:

I hope people already know about @bitmap’s script and mine to help with that.

Regarding Debussy, I think the low-hanging fruits (obvious related works) have already been dealt with, and what is left is mainly works existing with several common instrumentations (Clair de Lune, Petite Suite). Same with Ravel (Bolero, la Valse, Miroirs). So the recordings must be checked one by one to be linked to the right work (or to a “catchall” one)

I guess people would be interested in a script to set “[unknown]” for selected recordings on a release? I could probably do that in the next few days

I’d like to have some feedback on how much research is expected before giving up and setting [unknown] recording artist. My guess is:

  • if no information in edit history
  • if no info from the cover
  • if no acoustID or no useful data in acoustID
  • if no discID or no info in freedb with this discID (discid search)
  • if no info from the label website
    then setting [unknown] is justified. Did I miss anything, or is it already too much?

+1 for Debussy (catalogue of works nearly complete, only one opera) or Ravel. Brahms and Chopin for the same reasons, but they have much more recordings on them.

Count me in (but yes, later, please)

2 Likes

Yes, please! :slight_smile:

That list sounds about right, except that sometimes even if we only have track times a match can be found via AllMusic track duratons :slight_smile:

Hello. I am not quite sure about how “movements” and “parts” can be modelled in MB. When I list the (blastingly cool) list of Messian’s works, I see Turangalîla-Symphonie, I/29: I. Introduction and Turangalîla-Symphonie, I/29 as two separate works. Is there any way the second one is identified as “part of” the first one?

Sorry if I missed something. I’m quite new here :wink:

Both pages show the “part of” relation. “Turangalîla-Symphonie, I/29: I. Introduction” (your first example) has two relationships, “composer” and “part of”

2 Likes

I’d like to toss Carl Nielsen in there as a possible target. (Note also that there are a couple of Danish “Carl Nielsen”'s, so there might be some misattributions as well, maybe.)

Nielsen is basically done (less than one page of recordings) which seems a bit too little.

1 Like

Two more popular composers with a negligible operatic work: Felix Mendelssohn and Anton Bruckner.

The difficulty with Bruckner is that his symphonies exist in several versions, and sometimes finding the right work to use in relationship requires a long investigation. That being said, there are not many recordings left without related works…

Thanks for this info, didn’t know about that. That’s some obstacle of course and a reason why not to pick Bruckner at the first place.

Villa-Lobos is another possibility. @ListMyCDs.com, what would you say out of all the ideas that came up on the thread? :slight_smile: Maybe you can select two or three and we can vote on them

There was already some interest for Debussy so that’s a clear choice. Mendelssohn seems to be having decent amount of recordings and a lot of thing to do with works (if we don’t limit only to recording cleanup). Prokofiev was also mentioned and seems it would have decent amount of work to get it fixed.

On the poll for next month we should keep Prokofiev (5 persons voted for it). Any other names people would like to recommend? Some composers to consider could be Dvořák, Grieg or Mahler. These all have about 20-30 pages of recordings to clean. I personally don’t care that much who is selected. I believe we should keep selecting popular enough composers to get enough volunteers for cleaning.