๐ŸŽ‰ Happy 19th Birthday to Danbooru! ๐ŸŽ‰

Tag alias: *knee*

Posted under General

ghostrigger said:

create alias hands_on_knees -> hands_on_own_knees
create alias hand_on_knee -> hand_on_own_knee

Link to request

clarity and disambiguation; majority posts

-1. While this might seem like an obvious thing to support, it actually isn't. The reason for this is because of how messy the current aliases are in regards to these types of tags. The current aliases go all over the places, with some of the hand/s_on_<part> tags going to hand/s_on_own_<part> tags and others going to the hand/s_on_another's_<part> tags. This lack of consistency can only create problems, not resolve them. All these aliases should be fixed before trying to create any more of them.

The clarity and disambiguaty argument is questionable. The only reason for the necessity of it is due to the fact that we're not consistent with the naming scheme and aliases of these tags to begin with. If we were consistent and only used the on_<part> tags for self, then the existence of the on_another's_<part> tags then makes it clear the division of purpose between the two tag sets. In most cases images of characters touching themselves is way more common than touching someone else, so if we are just consistent people will easily figure out which is which. We won't have to start going crazy with these tags and end up reaching a point where we'll have to alias tags like hands_on_headphones to hands_on_own_headphones, just because of the minority images depicting characters touching someone's headphones as they're wearing them.

There are 3 exceptions where it is more common for depictions of touching someone else than of a character touching themselves. These exceptions are shoulders, head, and back. Shoulders because its more common to put your arm around someone else than touch your own, head because petting and putting a hand on someone else's head for fellatio or cunnilingus is common in images, and Back because like the other two there are common acts that involving putting your hand on someone else's back. For the rest though it is of images of the character touching themselves and not someone else that is way more common, so as long as we stayed consistent with the tag naming and removing confusing aliases the consistency of the system would keep users to use the appropriate tags.

imho, i don't see how things are getting messier and creating more problems. actually, the rise of clearer and more specific hand/s_on_own/another's tags should tell us that the older and more populated but ambiguous hand/s_on tags are getting a little harder to search on for specific posts. from whose hand/s and to whom and to which part/s? if we are in the olden days, it wouldn't be a problem. but, is our collection getting smaller?

now, if your talking about the precedent of the aliases, it just so happen, because the vast and undeniable majority are used that way. it would be extremely hard and impractical to point to something else. the greatly improved autocompletion should help and encourage old and new users to use the new and unambiguous tags. if anything near 1/2 or at least 1/3 of the population of a particular "on" tag is used for another else other than "own" or "another", i would ponder hard doing an alias.

our goal is not to make aliases of "on" to "own" or "another" but ultimately to discourage people from using the older tags and use the new and more specific equivalents.

these aliases we propose are for convenience and educate users that these older ones should no longer be used. hopefully, time will come that no one will use or create more the plain hand/s_on_part tags.

if we let the status quo, isn't that creating more problems in the long run?

You're not getting the point, many of these current aliases should be scrapped and unified to what they point to. The fact that some of them go to *_own_* tags and others go to *_another's_* tags just adds unnecessary confusion with their lack of consistency. For example having hands_on_knees -> hands_on_own_knees, but then we have aliases like hands_on_cheek -> hands_on_another's_cheeks. That will simply cause confusion when using autocomplete for the alias, as in some cases it points to own tags and in others it points to another's tags. The minimum is that all the current existing *_on_* aliases should only point to the *_own_* direction, since the "own" concept is the dominate form for the bulk of these tags, but preferably the _on_* tags shouldn't need to be aliased to anything else.

You seem to be confused at what an aliases is. An alias isn't erasing the term from existence on the site, it is saying that Term X is equal to Term Y. We have a major inconsistency by having some of the _on_ terms direct to _on_own_ tags in some cases and in other cases direct to _on_another's_ tags. You can't have the same terminology be aliased to two different competing terminologies.

Of course having both own and another is also in my opinion completely unnecessary, as the existence of just one of these types alone would define what the various *_on_* tags are if we had simply had consistency in usage. It was only lack of consistency in the earlier usage that forced having to slap on the unnecessary wording additions for both. If we had consistency with using _on_ to refer to self and then added the _another's_ for when touching someone else having to rename things to "own" would be completely unnecessary. Especially now with the autocomplete to show users that the two naming sets exist. As I already stated, going down a path where the naming scheme will force us to use hands_on_own_headphones instead of hands_on_headphones is ludicrous.

I'm not asking for the status quo, if anything I'm asking for improvement over the system that is currently in place and avoidance of using a name scheme that shouldn't be overused unnecessarily.


just exactly how many current aliases are we talking about inconsistencies? scrapping and unifying them is actually more harmful than good. we are talking about hundreds and even thousands of posts here. hard to find posts will be drowned down under the heap because one dominates the other.

where did exactly i said an alias being erased in existence? what we hopefully want to achieve is to discourage people from using ambiguous tags. to be more specific to particular point of interest. the alias is there to point what is more desired tag to use.

The minimum is that all the current existing *_on_* aliases should only point to the *_own_* direction, since the "own" concept is the dominate form for the bulk of these tags, but preferably the _on_* tags shouldn't need to be aliased to anything else.

reasonable. however, we have to consider how tags are used by the users. it will be equally confusing to point it to something else other than where it was used more often since time immemorial. we should not only focus on the semantics but more importantly how exactly how a particular tag is used. a few noise shouldn't prevent a legitimate alias just for the sake of being consistent for a remotely related tag. we should look on this on a case by case basis.

if you insist that such presence of a few tags being a "major inconsistency" then give it a time until nobody else or only a few people are using ambiguous tags. that users are introduced to new tags that should be used instead. these types of problems actually exist because people in the past tolerated such ambiguity or failed to foresee how the site would grow. if we follow the same attitude and inaction how exactly do we solve ambiguous tags so large. by creating new tags when it's too late?

There are currently around 10 aliases that should be scrapped (only representing 4 body parts), that represents some 5,479 posts. Those aliases direct various _on_<part> tags to _on_another's_<part> tags. The aliases should never have been made though and at the time they were done they should have been mass edits instead of aliases. To back that claim up all you have to do is look at our current *_on_<part>s tags, there are 15 currently existing with over 200 posts. Of those 15 only three of them are used for to refer to touching someone else. Those three represent around 5k posts (over estimation) and only 2 bodyparts, the head and the shoulders. The rest all represent a domination of the term being used to refer to a character touching themselves and represent 6 bodyparts/locations. They account for over 31,000 posts and that is after shaving off 10% of their post count to assume they were being used on images of touching another person. The honest value for using them on images of touching another person was likely far below even 5% usage and that was after checking sample populations or in some cases looking through all the posts for each tag. 31,000 posts versus 14,000 posts (I added in the slashed off 10%), it is clear which is the dominate usage for _on_<parts> terminology and that is not to refer to _on_another's_parts usage. So the existing aliases are incorrect in directing users to _on_another's_<part> tags.

You're essentially treating the aliases as erasing the terminologies existence, which you show with your own statement on how this will discourage using ambiguous tags. An alias as it functions is not correcting the user to use disambiguous tags its telling the user that the term _on_<part> is equal to the term _on_another's_<part> or in the proposed aliases _on_own_<part>. You can't say that _on_<part> equals _on_another's_<part> and then turn around and say _on_<part equals _on_own_<part> at the same time. It should only equal one of those terms, not both at the same time. You're not making things disambiguous, all you're accomplishing is making the _on_<part> terminology more ambiguous, which can be in turn used to justify further expansion of the _on_own_<parts> usages.

I am focusing on the tags usage. The vast bulk of _on_<parts> tags are specifically used in the manner that is the same as the _on_own_<part> tags (which is why imho the _on_own_<parts> tags is unnecessary naming). It is confusing to retain these aliases that follow a different ordering than the majority, particularly when the things aliased also follow the same domination of individuals touching themselves being more common form than of those touching another person. The aliased tags represent hand/s on checks, chest, face, and stomach. For all 4 of these categories the *_on_own_<parts> version is the more populated version than _on_another's_<parts> version that the aliases direct to. So the aliases are really following an incorrect usage in comparison to the majority of the tags.

The problem is consistency and the pushing of the unnecessary _on_own_<part> tagging naming scheme. These aliases fail with consistency no matter how you look at them, and without things like that muddying up the water the necessity of having to use the _on_own_<part> naming scheme falls by the way side as well. The only thing needed was the _on_another's_<part> set of tags to go with the _on_<part> naming scheme and there really would have been no issue.

if you are personally against the current aliases trend and proposing instead radical mass edits, i think it would be desirable to create a separate thread for that.

your argument to give -1 for aliases ( hands_on_knees -> hands_on_own_knees and hand_on_knee -> hand_on_own_knee) because hands_on_cheek -> hands_on_another's_cheeks exists is ridiculous. we are going off topic, but the semantics can be discussed separately.

ask yourself how many of the posts in hands_on_knees are used other than hands_on_own_knees? how many hand_on_knee are used other than hand_on_own_knee? these older tags are used for how many years? how many incorrect posts do you see? how many posts are considered noise? by how much? 10%? 20%? 50%? isn't an alias both logical and viable for these?

if you want, you can prepare the list of tags that you claim "major inconsistencies" in the site on a separate thread. so we can discuss thoroughly on each case and not just throwing numbers so other people who have no time unearthing the tag list can judge for themselves if these aliases are harmful as you claim.

It's not off-topic, you're just not getting the problem with your proposed aliases. Even without the whole hands_on_own_<part> naming scheme being horribly unnecessary, going by the current (bad) aliases that exist, your proposed aliases are bad because they go in a direction opposite of the existing aliases. Either your proposed aliases are wrong or the currently existing ones are wrong (and I'm of the opinion they're both wrong). Let me break it down for you.

  • Your 2 proposed aliases:
    • hands_on_knees -> hands_on_own_knees
    • hand_on_knee -> hand_on_own_knee
  • All other existing hands_on_<parts> aliases that exist:
    • hands_on_cheek -> hands_on_another's_cheeks
    • hands_on_cheeks -> hands_on_another's_cheeks
    • hands_on_chest -> hands_on_another's_chest
    • hands_on_face -> hands_on_another's_face
    • hand_on_belly -> hand_on_another's_stomach
    • hand_on_cheek -> hand_on_another's_cheek
    • hand_on_chest -> hand_on_another's_chest
    • hand_on_chin -> hand_on_another's_chin
    • hand_on_face -> hand_on_another's_face
    • hand_on_stomach -> hand_on_another's_stomach

Your proposed aliases would be the only existing aliases that point in a direction different from the existing similar aliases, and as I said either your aliases are pointing in the wrong direction (ie they should be pointing toward hands_on_another's_knee and hand_on_another's_knee) or the existing aliases are wrong and should be scraped.

The existing aliases are wrong. 12 out of the 15 hands_on_<part> tags (those with 200 or more posts) are used specifically for images of characters touching themselves. Similarly the tags that are aliased follow this same pattern, with the more common depictions being the hand_on_own_<part> tag and not the current direction they point to. From looking at the currently existing tags, outside of the tags for hands on ass, everything is 90% and more of one specific type of depiction. For 12 out of the 15 that depiction is hands on themselves, for the remaining 3 (which covers only 2 body parts) it is hands on someone else. In short hand/s_on_<part> tags are used more often to refer to the same concept as hand/s_on_own_<part> tags, than they are to refer to hand/s_on_another's_<part>. The fact that we have these hand/s_on_<part> tags aliased to hand/s_on_another's_<part> is inconsistent with the usage of the terminology.

Your aliases are unnecessary, at least in the direction you're making them go. It was never necessary to go down the route of having to use the hand/s_on_own_<part> naming scheme. The mistake of the past was and still is a failure of standardization and consistency. You argue that pushing all the hand/s_on_<part> tags to hands/s_on_own_<part> naming is necessary for removing ambiguity, but in usage the tags never really had that problem. Your own example that you use is one used out of failure to look at how the tag is used and throwing out wild assumptions. You argue that there is a problem with hands_on_knees (and lets throw in hand_on_knee as well). You then throw out numbers like 10%, 20%, 30% or more in potential inconsistency in usage. The actual numbers? Only 8 out of 719 posts for Knee_on_hand is used to refer to touching someone else, and as far as I could tell 0 posts out of 780 for hands_on_knees when I checked that tag. That's 1% and 0% respectively. There never was a major problem, it was always a false assumption that there was and that the naming scheme was necessary to solve this problem that actually doesn't exist.


imho, honestly i think you are a bit lost and a bit obsessed in shutting down legitimate alias requests for the sake of "consistency". let me remind you a fellow mod had just spoken there's no such thing as hard rule or policy in danbooru, only guidelines. you're pulling a "consistency" card because you're claiming that such deviation is a "major inconsistency", "messier", "confusing", and "unnecessary".

the final tag aliases (or the proposals at the very least) should be evaluated based on tag usage, history, clarity, familiarity with the userbase, and population size. it should stand ground on its own merit, not because of an unrelated tag or remotely similar one that happens to be syntactically the same.

contrary to your claim that i'm erasing the existence of a tag, older tags retain their lives in aliases (at least in this proposal). that's why factors like history and tag usage play a very important role to consider if they're even alias-worthy or just tag move/mass edit would suffice. so users who used the tags since forever will have no time trouble migrating. tags change in time. with the autocompletion, they're notified or made aware that there's a more desirable tag to use now. and it's preferable to use that new tag. in the absence of a reliable wiki from both which hands_on_knees and hand_on_knee sorely suffer, where do we look into? how exactly were they used in the past? the noise factor by how much? we are talking about 700+ posts used nearly half a decade. despite the lack of a wiki, were they used consistently for a single thing? mistags by how much, just in case?

for new aliases which existing older equivalents, history is very important. we have to balance the older and new tag users. consider also that not many people are forum participants. we can't just "kill-off" a tag which have been actively and consistently used only a for single thing (with considerable population to boot) as same as a newer tag and expect people not to use the former. leaving it blank or empty, wouldn't solve that it's a mistag and taggers aren't informed of a new and better tag.

the surviving general tag in an alias should be specific and should not inherit the ambiguous nature of its predecessor, if ever that was the case. both the new hands_on_own_knees and hand_on_own_knee confirm this and retains the autonomous nature of each other as discussed 5 years ago in this thread. i don't see why you want unifying "to what they point to" now.

it is worthy to note, that the older but ambiguous tags of interests came into existence independently of each other. they didn't copy the parameters or preconditions of the other. it might not be obvious to you, but they just rose out of need to look for a particular post. on one tag, a character's hand is more common on this and that and so on, but not on the other tags. as time goes on, complexity in searches arises. the newer but more specific and longer tags came because they resolve the questions of whose part, to whom, and to where all in a single tag where not even a combination of multiple older tags can duplicate. people are asking, what tag do we use for this or for that? you don't have to believe me, but just search the forum and note the users (and user levels if you want) how people find convenient the new tag names of own/another's which you describe as "completely unnecessary", and "horribly unnecessary". go on, that's your sentiment perhaps some others too, but others (on record) find it helpful for searches, clarity, and specificity - direct to the point. i won't be the final arbiter here. but i'll respect the decision, whatever that will be. however, i'll continue to advocate a clearer and unambiguous tags that you find "horribly unnecessary".

you are claiming -1 for alias citing "consistency" issues because of own/another in the opposite direction. ironically, you are tolerating inconsistency by retaining the ambiguous hands_on_knees and hand_on_knee tags despite the trending clearer tags. the keyword here is to be specific, to give clarity. in your own words, what you really find "unnecessary" is actually necessary because of consistency, not because of syntax and semantics of the "own/another's" thing, but the need to be clear. it doesn't matter if it does the job even in its ambiguous form now. time will come that counterparts, will be harder to find as population grows. heck, counterparts aren't even required if they don't even exist yet. the newer own/another's tags are scalable enough for future situations. if you have a better solution, i'm all ears.

as i said, you are off-topic, if you want changes on how existing hands_on_<parts> bump their respective threads. you claim "the existing aliases are wrong" and you did nothing about it (considering they are actively being used for some time now), and raised it only now, in a knees thread? something so serious should be on a separate thread. different situations call for different measures. also review their histories why exactly they came into being and approved. you're complaining to me, a mere user, why these aliases were approved? you have the power and you knew better, you know who to direct your grievances.

" argue that pushing all the hand/s_on_<part> tags to hands/s_on_own_<part> naming..." - never said this.

our goal is not to make aliases of "on" to "own" or "another" but ultimately to discourage people from using the older tags and use the new and more specific equivalents.

The aliases do have value because they show up in autocomplete and it becomes clear what the meaning is. In this case it makes sense for X_on_Y to map to the most common version. With that standard many of the existing aliases need to be fixed.

There are enough posts that distinguishing between own/another is useful so it doesn't make sense to alias both of them to X_on_Y.

We could remove all the aliases and clean up the tags, but I have a strong suspicion X_on_Y would creep back into usage.

I may end up regretting the precedent this sets.

albert said:

There are enough posts that distinguishing between own/another is useful so it doesn't make sense to alias both of them to X_on_Y.

I never said both need to go to X_on_Y, just that X_on_Y having to be moved to X_on_own_Y is unnecessary if we just used X_on_Y alongside X_on_another's_Y. The presence of just the one defines the purpose of the other, and consitency of usage would maintain proper usage. Current usage shows this to be true with nearly all the remaining X_on_Y tags. The step of converting over X_on_Y to X_on_own_Y was actually never necessary outside of imagined fears, as shown by ghosts's own words in assuming significantly higher misuse of the tags than actually are occurring.

The only thing needed was migrating the few cases where X_on_Y is being used for X_on_another's_Y, and then unifying the usages of X_on_Y tags.