Disambiguations: a collection of examples and discussion of needs

Tags: #<Tag:0x00007f0509afc700> #<Tag:0x00007f0509afc548>


The guidelines say not to store data in the disambiguation. Using data, that has already been entered into it’s correct MB field, in the disambiguation field to provide much needed context would not be “storing data in the disambiguation”.

I think there is a tension between recorded music encyclopedia users who want reliable visible context and database focussed people who want order and clearly defined relationships.


I definitely agree. All I meant was that, of all the types of entities, works are the only one that I can see supporting any weak argument for not including disambiguations. Could have worded it better, though, sorry.


The work guidelines says to use disambiguations when they seem useful, but was added to avoid people adding “Scorpions song” disambiguations to every single Scorpions song and the like (since that data is shown both in normal and inline search), not to heavily restrict disambiguations otherwise.

My general opinion is that disambiguations should be a last-effort thing, and we should just display the data we have better when searching, mostly because disambiguations aren’t translatable. At least unless we decide to do like Wikidata and have a different disambiguation field for every language!

That said, until that data is displayed better (like it generally is for works), it does make sense to have disambiguations on anything that has any risk of being confused. I probably still wouldn’t add it to an artist called “Fissfhearistofeles” because it seems very unlikely two people would come up with the same nonsense name anyway, but I’d say it’s reasonable to add it to anyone with a standard English name, for example.


Speaking of this, I remembered that one of my old edits adding disambiguation was even voted down and therefore failed: https://musicbrainz.org/edit/19080916. It’s currently one of the only 4 edits of mine that have been voted down so far. The artist appeared in the Duplicate Artists report because its sort name/alias “Rebecca” matched one of my artists I was subscribed to at that time. Even though I mentioned that I would redo the edit again in the future, I never did because I was afraid I’d just get voted down again (and currently it isn’t bothering me anymore since I don’t have the other artist in my subscriptions anymore).


[quote=“culinko, post:24, topic:249806”]
I remembered that one of my old edits adding disambiguation was even voted down and therefore failed[/quote]
It’s very easy for a single opinionated voter to fail edits, since other people don’t know about it.
If you bring contested votes to everyone’s attention via the forums or similar you’ll get a much more… democratic result :confused:


[quote=“aerozol, post:17, topic:249806”]

I agree with aero here, in fact when rafwuk wrote about “going back and removing the ones I’d already done” I actually audibly said “NO” out in the room.
so horrible, loss of data! :/p
Myself I follow the guide of usually filling in the disambiguation for every artist I enter. I usually do the same things others have mentioned in this thread, adapting it to what is logical, and what I know/found out about that artist.

I try to be as generic as possible and considering who the audience of this disambiguation is for; they know nothing about this artist.
So for your Randy Jackson example, I’d instead write years active, place from (american) and include the (possibly) most famous band he was in: the Jackson Five, instead of saying “he is the brother of these other people” since that is idk, rude or something to him as an artist, he shouldn’t be best known for who he is related to or married to, but what he himself has done and what group(s) he has been in.


Just a little example:
I added a compilation album with a song from Ed Sheeran.
On the tracklist I pressed Edit for the artist and there are actually 2 entries:

With this simple disambigaution “famous” and “unknown” it’s clear for me, which one I have to choose. It may be frustrating for the “unknown singer-songwriter” to see this disambiguation, but for an editor its a simple and clear distinction.


I don’t know any Ed Sheeran and I think I would prefer some era of activity, country, music genre or a combination of that.
Probably one of them was active in 19XX and the other in 20XX.


You never heard about https://en.wikipedia.org/wiki/Ed_Sheeran and https://en.wikipedia.org/wiki/Ed_Sheeran_discography?
How can it help to know that he wrote three #1 albums for the UK-Charts, two #1 albums for US-Charts in 2011, 2014 and 2017 if you don’t know him? :innocent::wink:


I’m with @jesus2099 here. And I probably don’t want to know either. :stuck_out_tongue:


I’m not saying he sucks, he’s just not on my radar.
And in any way, for any artist, we should stay factual but still helpful in comments:


You are right.
But this can become very difficult:


I only referred to your case with famous artist (for some people) versus unknown artist (for some people), which is not very appropriate, IMO.
BTW I don’t know what does not apply to Aura in what I said.


Example: I’m searching for the correct house/trance band “Aura”. Which one would be correct of this 5 (from about 30) Aura-variants:

  • Aura (Trance; René Ubachs; Released Air / Energy Transepose / Destination Unknown)
  • Aura (house, released Share the Tears / The Chant)
  • Aura (trance, released Intercity / Walk on Earth)
  • Aura (ambient house, released Communication Native)
  • Aura (Dance/Trance (released You Came, When I See You Again))

In every case I have to know what they already have released.
Does it not lead to the fact that you have to search for a single recording anyway and hope that someone else has already added the correct person/group to the song you want to add?

What I want to say:
At least in the tracklist editor a popup/mouseover with a list of artists with similar names for the same song would be more helpful then the search result I posted above.


But editors adding a disambiguation don’t always know much about the artist. For example, usual dance music compilations with tons of obscure artists… sometimes it is very hard to add a useful info, just because you can’t find any.

I don’t see where is the problem, because if one has a better disambiguation he can just replace the existing one.


This No vote was a bit harsh imho.

Especially for this artist, from annotation (by the editor who voted no):
Traditional Chinese: 許同恩
Simplified Chinese: 许同恩
Wades-Giles romanization: Hsu Tung-En
Hanyu Pinyin romanization: Xu Tong En
English name: Rebecca Hsu
Performance name: 同恩 / Rebecca

This artist has apparently multiple names, “Female pop singer from Taiwan” is much more useful for me than "Hanyu Pinyin romanization: Xu Tong En"
By the way, is she the same person on https://www.facebook.com/同恩-Rebecca-196103839114/ ?


Between disambiguation and annotation, there really is no reason for us not to know who some of these people are.
I mean, I have left annotations on Eric Darken and Eric A. Darken asking if anyone could confirm that they are separate people, simply because I can’t find enough detail on “Eric A” to know.

All it takes is a simple “17th century composer” to let people know that it is not a twenty year old DJ.

  • I just sent a message to Eric Darken asking if he is Eric A. So, with any luck, that will be another mystery solved.


A possible solution is to have the software to automatically generate a disambiguation from the data we have. We can say an artist is mainly a composer because she has more work composition relationships compared to other types of relationship. Combining other personal details, we can produce “composer based in Finland (1933-2001)”. Users can of course override the automatically generated disambiguation and are encouraged to do so. The automation just prevents editors from creating false relationships where there is no disambiguation.


And we should definitely do this for live recordings, since the style guidelines specify that these recordings should have specific disambiguation comments, not to mention also the benefits relating to i18n. Years ago there was a bot who automatically generate disambiguation comments using relationship data, but I think that bot has retired.