Disambiguations: a collection of examples and discussion of needs

Tags: #<Tag:0x00007f2a0456eb88> #<Tag:0x00007f2a0456e958>


I often find myself editing releases in genres that I know next to nothing about. Disambiguations can be very helpful here when they guide me quickly to the correct choice. But the usefulness depends on the disambiguation having an association with something I know already.
The example from the Documention illustrates this well,
Randy Jackson (brother of Michael and Janet).
Great if you know who Michael and Janet are.
Otherwise no use.
As the database grows there will be more and more duplicate names. Maybe the approach to disambiguations needs looking at - what situations are they most critical in, what content is needed in the disambig, who the target audience is, where they are succeeding and where they fail.
So I’ve started a list to collect some instances so that we can get an overview.
Please add real-life examples and comments that you see as contributing to a better understanding of efficient disambiguation.

[STYLE-792] Additional guidelines for disambiguation comments

One thing that I’ve found problematic is not putting disambiguation on entities that don’t have duplicates yet. Especially with individuals, when there’s very minimal info available. Putting a disambig reflecting something of the source of the info would be helpful when another person (like me) comes along to enter someone with the same or very similar name. Yeah, there might be plenty of artists for whom that never happens, but it takes way more time to dig up info on an artist without a disambig than to type in a few extra words.


I wouldn’t (and don’t!) worry about that. It’s better to have a uniquely named artist with a disambiguation than three artists with the same name combined in the same MB:Artist because there was no disambiguation to start with, is my take.


I think the UI could also do a better job serving information we have about artists in the search results. If someone is credited primarily as a producer or player of a certain instrument, why isn’t that information visible when I’m searching for an artist to add to this relationship I’m editing? Likewise with other relationships. We wouldn’t need the disambiguation comment ‘member of So-and-So Band’ if the search screen just told me that based on the artist’s relationships.


Presumably (particularly true in this case), this Randy Jackson is not famous enough on his own, or not as famous as Michael and Janet are. If you don’t know any of them, this is probably not the Randy Jackson you are looking for.

At least, it points that you have to check for homonyms before crediting Randy Jackson. It turns to be more useful when you consider disambiguation comments for other Randy Jackson artists at the same time.


For someone editting in 1970s American pop music, with no familiarity with the genre, excluding "Randy Jackson (brother of Janet and Micheal) "
on the basis of not recognising Michael and Janet could easily be a mistake.

I think this goes to who the target audience of the disambiguation is.
Using Randy as an example:
If the target is afficionados of late 20th century American pop then it is a useful disambig.
The numbers of such afficionados are probably decreasing already however.
And, if MB is global in aim, such afficionados are probably already a minority.
So who is the disambig aimed at?
To answer that question, knowing when the disambigs are being used, and in what context, seems necessary.


I think that many disambigs could be superceded by having ready (mouse-over?) access to metadata on artist/work etc.
I find that I’m often putting instrument/profession, birth/death dates, geographical area and/or genre into disambigs.
eg John Bowie (1759-1815, UK, publisher, musician, cataloguer)
All of which is redundant in content but useful in many current GUI contexts.
Genre appears to be the common unique data.


Somewhat tangential:
I’ve added disambiguations to the first half of my list of Subscribed Artists.
That section of my list now appears to have a lot to do with music.
The immediate relevance to musicof the non-disambiguated areas, is much less obvious.
I then compared the “contextualised” list with the visual layout/design on a random artists page.
And am now thinking that, while the layout is nice and clean, which is great for editting, a naive music-loving user might find it somewhat clinical, “machine-like”. Which might not fit with your average music encyclopedia user’s aesthetic.

Don’t get me wrong, when someone described the MP3Tag GUI as “a homage to Excel” I thought, “Yes, isn’t that good”.
And I do think the uncluttered design is great for editting in.


Disambigs are great for now, I put one in for every artist I create, but almost everything I write could be automated.
Country/area + top genre tag/s + group or individual
A mouse-over could add date ranges as well, but having it all in a table right away could be useful. Then you could arrange by different criteria as well.

I don’t really have anything too useful to add though, disambigs are great, and let’s vote for:
To get on the road to using proper genre tags (or entities) :wink:


For search, showing things like area and start/end dates seems to make sense. Possibly even for inline/drop-down search, although screen real estate is at a premium there.

But things like “producer”, “flautist”, etc are far less obvious to distill automatically (especially taking localisation into account), especially for artists with dozens or hundreds of ARs. Same for things like memberships in groups.

In the inline context, if the dropdown list could include an icon that when clicked opened that entity’s page in a new window, that would be convenient to check on multiple ambiguous hits.


I don’t really see how that could be hard. You just pick the most or two most-used relationships to display. It would require one-word names for each relationship though, like indeed ‘producer’ and ‘flautist’ including localisations, but that is a small price to pay. Most instrument relationships don’t have very specific names luckily, it is usually ‘<<>instrument name> player’, so only the exceptions like lutist and organist need to be defined and specifically translated. It is not as much work as it sounds.


Sorry to be late to the party on this one but, I have a standard I use here for disambiguation that works pretty well for me and I’ve spent a lot of time splitting out artists here.

5 points of information usually in this order

  1. Era
  2. Location the artist is associated with (country/well known city.etc)
  3. Genre
    4a. - For groups: grouping details (don’t know how to name this) - band, or duo or trio or choir or collective or group (what fits best with those)
    4b - for individuals primary roles/instruments anything from vocals or specific instrument or producer or engineer or graphics etc.
    5c - wild card - if the artist has a specific noteriety or signature song or label, or real name, or notable band member or alternate name or person they worked with
    00s UK post-rock trio "signature song"
    60s-70s Manhattan folk singer/songwriter

A major factor is that you only include information from as many number as you know, the rest can be filled in as required or known

When you have 2 remarkably similar artist (where more than 1 of their numbers are known, it may be best to focus on just the numbers where they differ.

I hope you can understand my ramblings here that are perfectly sane in my head and a lot easier to do than explain i’m afraid, but its about time i start writing some of this down.


Many artists are also associated with a certain label, so that might be another useful info to display, e.g. https://musicbrainz.org/artist/9cf27403-fbde-4bab-ba01-d05e879d2fe5.


Yes I like that as well, in fact I should have expanded on my point 5 and will include this


When I first started here I always added disambiguation comments when adding artists, my thinking being that it’s easier to do this now when I’ve just done the research, than have to find the information later. However I was told not to, because the field should only be used when necessary for disambiguation (and I then went around removing them from artists I’d added), and since then have only used them when necessary. The discussion here suggests that others agree it would be good to always use them. The guidelines are not clear on when to use them, and in fact might deter people from adding some of this information.


It does not matter how clear the disambiguation comments are, some people will ignore them. People frequently add compilations to this Various Artists, even with the disambiguation comment stating “German techno producer Torsten Pröfrock, DO NOT add compilations”.

I don’t know how the UI orders artists, it doesn’t seem to be alphabetical. If it were possible to weight the selection based on the popularity of the artist (for example, how many relationships this artist has in the database), maybe people would be more inclined to pick the one that’s most likely to be correct - and if it’s not the right one in a particular instance, there will be more eyes (and brainz) to fix it.


I was told this too, and I’m glad I ignored it. It still makes no sense to me.


I can kind of see it with works, since those are always displayed along with their writers, but disambiguations definitely help with just about everything else. Even if nothing else has a similar name when it’s added, it might come into play in the future. Some little-known artist marked “(singer)” is probably not going to be the one you want to add to a “photographer” relationship.


This would be great!

Also, searching the artist name as well as the title when it comes to things like recordings and works would be really helpful. There have been many times when I created a new work or recording because there were too many similar ones and I didn’t find what I was looking for even though it was there…


Works benefit from disambiguation often.
The most obvious need being for the catch-all for the work. The year and genre being other helpful possibilities. I am arguing against assuming that encyclopedia users or editors are knowledgeable in the area that they are reading/working in. I often don’t recognise the names of a work’s creator. Disambiguation can provide much needed context.