Topic: [Feature] Automatically set category of tags ending in *_(disambiguation) to invalid

Posted under Site Bug Reports & Feature Requests

Requested feature overview description.

When a new tag is created whose name matches the pattern *_(disambiguation), change its category to invalid. Prevent tags which match that pattern from having their category changed to anything but invalid.

Why would it be useful?

Disambiguating a tag is currently a somewhat fraught process. A two-step BUR is required, one to actually make the alias and a second to change the category of the target. Sometimes this can backfire due to the non-deterministic nature of BURs, requiring manual intervention: see topic #33446. I'm not sure exactly how the current system enforcing the lore category on *_(lore) tags works but it might be possible to reuse some of that. There will never be a valid reason to create a tag matching this pattern.

Vandalism potential is minimal since there's already nothing stopping people from applying disambiguation tags manually, this just gives them a different colour.

What part(s) of the site page(s) are affected?

Post editing, aliases/implications/BURs.

I worry that if we do it all at once invtags:>0 will become completely impossible to manage. It does need to be done eventually though.

I've also noticed that some users tag (disambiguation) tags manually since they don't have any aliases to them, what's up with that??

+1

What about disambiguation tags set to the character category? Common names which can only be characters (and maybe artists), like felix_(disambiguation), or jason_(disambiguation) are getting this treatment. https://e621.net/tags?commit=Search&search%5Bcategory%5D=4&search%5Bhide_empty%5D=no&search%5Bname_matches%5D=%2A_%28disambiguation%29&search%5Border%5D=count

There are a bunch of them, and this isn't something an automated system would be able to keep.

earlopain said:
What about disambiguation tags set to the character category?

They should be set to the Invalid category instead. There was an approved BUR to disambiguate and set many ambiguous names to Invalid, but not all of them took because of execution ordering, and new ones have been made since that didn't get set to the Invalid category.

earlopain said:
What about disambiguation tags set to the character category? Common names which can only be characters (and maybe artists), like felix_(disambiguation), or jason_(disambiguation) are getting this treatment. https://e621.net/tags?commit=Search&search%5Bcategory%5D=4&search%5Bhide_empty%5D=no&search%5Bname_matches%5D=%2A_%28disambiguation%29&search%5Border%5D=count

There are a bunch of them, and this isn't something an automated system would be able to keep.

Why not? There's no reason for them to be character tags. As you point out, they could also be artists, and some like biscuit or amber could even be general.

Besides which, the point of an ambiguous character name tag is that it should be changed. We shouldn't encourage people to leave incomplete character tags on their posts by showing them in the "right" category. The disambiguation suffix itself only exists as a public shaming tool, and it seems it alone isn't getting the job done.

Hell, I used a disambiguation tag because it was the most correct tag I could find. :(
I'll have to look for example. There really needs to be an "Itagged" keyword for search, sigh...

:edit: Oh, right!
Found it: https://e621.net/post_versions?search%5Bupdater_id%5D=27213
bow_(disambiguation) -> This should have different kinds like the French 'arc' aka like for arrows, a bow like ribbons in hair, etc. I'm not sure what keyword would work better for 'bows'. :(

Note: That image example is on blacklist for most folks. Blacklists works for direct links?

:edit2: I really hate to do exponentially expensive searches like this, but this found some tags. I'll fix that ambiguous tag. Sorry for messing that one up. I guess I was too tired to think of doing a hard search for the tag.
https://e621.net/tags?search%5Bname_matches%5D=*bow*

Updated

watsit said:
I can't help but notice you added male_(lore) to posts that were already clearly male. Lore tags should not be used when the proper non-lore tag is already used for the character (e.g. a solo+male post should not have male_(lore); only if the male is tagged ambiguous_gender or something other than male should male_(lore) be tagged).

Yeah, I'm going through my list for that kind of idiocy. :( Is it possible it was for a second character who wasn't clearly male? Sigh... debugging is far more painful than creating the problem in the first place.

OK, step 1) Fix that bow tag. Step 2) Go through and find those conflicting tags.

https://e621.net/posts/2723779 Shouldn't this one be ambiguous_gender? For ferals, it's kind of hard to tell male/female from that angle.
https://e621.net/posts/2764072 And this one... sigh.

:edit: Done except for the ones that might need ambiguous_gender tag - I should probably clean up the bow_disambiguation tag in general, too? Also, the other disambiguation tag. At least now I know how to find all my mistakes!

Updated

Possible to have a new tag category of disambiguate, or disambiguation?

(I just have nightmare of newbie(s) seeing a tag in invalid category and deleting it from posts because is "invalid")

listerthesquirrel said:
Possible to have a new tag category of disambiguate, or disambiguation?

(I just have nightmare of newbie(s) seeing a tag in invalid category and deleting it from posts because is "invalid")

The tags with the _(disambiguation) suffix should also be removed after adding a fitting tag for them. So, I don't see a reason to make an extra category for it.

Some of them are a pain because they're character names, and it's not 100% obvious since often they're OCs. XD

alphamule said:
Some of them are a pain because they're character names, and it's not 100% obvious since often they're OCs. XD

This is my biggest problem with these, I'm not sure how I'm supposed to work out what to do with a firstname_(disambiguation) tag that's been there for 10 years, no source or description, etc. I feel like the only option is to just remove it but I don't want to get into trouble for removing a disambiguation tag without replacing it with anything.

If it's only 1 or 2 posts, I am not worried. If it's 50 posts, then I need to yell at some uploaders, except they're probably no longer on the site. XD

In other words: Treat them like weeds. Don't let them get big.

dubsthefox said:
The tags with the _(disambiguation) suffix should also be removed after adding a fitting tag for them. So, I don't see a reason to make an extra category for it.

Decided to do a little look at tag changes.
Example: removal of sam_(disambiguation)
https://e621.net/post_versions?commit=Search&search%5Bexclude_uploads%5D=0&search%5Btags_removed%5D=sam_%28disambiguation%29

Various users have removed sam_(disambiguation) tag, and most of those edits are NOT accompanied by the addition of another sam* tag.

54 total removals of sam_(disambiguation).
9 edits that include adding another sam* tag.
54-9 = 45 tag edits where sam_(disambiguation) tag REMOVED and NO other sam* tag is added.

(only one of the edits included a "Reason", and it was just "Removed disambig tag.")

Seems like some users see a tag in invalid category, and just remove that tag for being "invalid".

listerthesquirrel said:
Various users have removed sam_(disambiguation) tag, and most of those edits are NOT accompanied by the addition of another sam* tag.

54 total removals of sam_(disambiguation).
9 edits that include adding another sam* tag.
54-9 = 45 tag edits where sam_(disambiguation) tag REMOVED and NO other sam* tag is added.

(only one of the edits included a "Reason", and it was just "Removed disambig tag.")

Seems like some users see a tag in invalid category, and just remove that tag for being "invalid".

Hmm, doesn't it automatically remove invalid tags, when editing a post? And if removing without replacing, that sounds like a bad idea. :(

listerthesquirrel said:
Decided to do a little look at tag changes.
Example: removal of sam_(disambiguation)
https://e621.net/post_versions?commit=Search&search%5Bexclude_uploads%5D=0&search%5Btags_removed%5D=sam_%28disambiguation%29

Various users have removed sam_(disambiguation) tag, and most of those edits are NOT accompanied by the addition of another sam* tag.

54 total removals of sam_(disambiguation).
9 edits that include adding another sam* tag.
54-9 = 45 tag edits where sam_(disambiguation) tag REMOVED and NO other sam* tag is added.

(only one of the edits included a "Reason", and it was just "Removed disambig tag.")

Seems like some users see a tag in invalid category, and just remove that tag for being "invalid".

Most of those edits were done by Kemonophonic, who has Privileged rank and really should know better.

For example, post #3390564 was uploaded by its own artist, and says right there in the description, "The fox is Sam and it's an oc of mine." Open and shut case for creating a new character tag suffixed with the artist name. But nope, blanket removal is an easier way to boost your edit count.

I have to take some blame on me too.

  • While uploading, I have discovered that the alias to invalid_tag has quite a high number.
  • My thought process was:"if a post has a sufficient amount of tags, invalid_tag can be removed"
  • I went to the Discord help-desk and asked:

"The invalid_tag is getting out of hand. Can I go through the first few pages of invalid_tag order:tagcount with the tag script, to remove it from posts with a sufficient number of tags?"
Answer by a janitor:
"I would say that to be fine, invalid tag is one of those things where they should be moved into invalid category instead, but would need to be done by admin and there's still nobody really handling aliases and implications to my knowledge outside of few if they happen to have time"
Me:
"ok. if no one interrupts me, I am going to do that for every post with more than 30 tags"
Janitor:
"yeah, this is one of those fucking cases where technically you shouldn't, but if you don't then the problem continues to exsist for years, so at least you now have someone to point finger at if you get record for some reason"
Me:
"and I mean... I asked in the help desk"

  • I go to the search page, search for invalid_tag tagcount:>30 and apply -invalid_tag with the tag script.
  • I kept going with my daily business until the night.
  • Plot twist: If you search for invalid_tag, it searches for all tags in the invalid category. And it also removes all of them if you apply -invalid_tag. What is deviant from all other tags.

that's not true. I think I misinterpreted the post changes page. I guess my problem isn't as big as I thought

  • I noticed my mistake and messaged an admin about it. He said I should just keep it like that, and don't have to reverse it.
  • But being reminded that character tags are also in the invalid category, make me overthink that again. I'll reverse that...

Updated

dubsthefox said:

  • My thought process was:"if a post has a sufficient amount of tags, invalid_tag can be removed"

They can be. invalid_tag is supposed to be a tag that can be removed on sight. If possible, you should also try to fix up any other bad tagging you see on such a post, but the purpose of invalid_tag is to delete it when possible (as opposed to other Invalid tags that should be fixed). Please don't add them, the tag is hard enough to manage as it is.

I figured out undoing soft, for instance, just adds back invalid_tag. So I just added invalid_tag to about 200 posts... I am going to toaster bath, I guess...

Edit: It doesn't remove character disambiguations because they are not in the invalid category... I don't have to toaster bath... still stupid move by me, sorry.

Updated

dubsthefox said:
I figured out undoing soft, for instance, just adds back invalid_tag. So I just added invalid_tag to about 200 posts... I am going to toaster bath, I guess...

Mistakes happen.

But I would be remiss to not bring up an idea I got from reading wat8548's post. Make it so you can't add invalid tags to a post when editing them. It's one thing for an invalid tag to mistakenly be added when uploading, so the uploader can know to fix it without delaying the upload, or as a result of a later alias, but I can't think of a reason for an invalid tag to be added as a result of a tag edit. If you know the proper tag, you should add that tag, but there's no point in tagging, e.g. bob (aliased to bob_(disambiguation)) on an existing post if you don't know which Bob it's referring to. The post existed without the invalid tag already, and anyone who would later tag the characters would still need to figure out and verify who the character is anyway.

watsit said:
But I would be remiss to not bring up an idea I got from reading wat8548's post. Make it so you can't add invalid tags to a post when editing them. It's one thing for an invalid tag to mistakenly be added when uploading, so the uploader can know to fix it without delaying the upload, or as a result of a later alias, but I can't think of a reason for an invalid tag to be added as a result of a tag edit.

I would say the opposite, actually. Ideally we should block both, but the vast majority of invalid tags are added by users who have never touched the edit button in their life. In fact, it's weirdly common for users to not even know there is an edit button.

And as for an example of when adding an invalid tag might be the right thing to do...

  • 1