[X] My Assistant

I've been using this site for ages and love the tagging system. I honestly haven't read the 50 pages previous to this one on the feature requests, so I don't know if there is much related to this.

Feature improvement: I've been thinking how nice it would be to implement tagging per image so as you go along and see something tag worthy you could enter it. This per image tagging would appear at the very bottom of page and would be the normal text field for tagging. I do understand that tags are meant for typically a majority of the content ex. should contain x number or % of pictures before being tagged with subject matter. For that you could implement low power tags or something along the lines that make it visible to users to vote up or down but not searchable until a certain levels. This could be all the change if you wanted to keep it like that or in addition you could make it so each page tagged can be found by clicking on the tag and having it list all the pages that tag has been found in that particular gallery ex. pages 23, 204, 1043 I'm also thinking it would help the tagging stem be able to say yes this tag appears on all of these pages there for is more or less valid.
Reason: Allows users to find content they're looking for faster especially in galleries with 100's of pictures. Doing it this way also gives every user going along a better incentive to tag that pages items and lets those reviewing tags the ability to find what page(s) tags came from to find tags that are untrue or that since every most pages contain tagged item that it should be voted up.

I wanted to get other peoples feedback on this specifically the Almighty Tenboro who's hard work has made this site what it is Thanks..

Sorry about this long dribbling feature request

https://e-hentai.org/g/541034/1444043f65/
There's my random gallery, a mere 232 images right now. Now open a text document and use it to "tag" that gallery image by image and see how long that takes. Remember that this is just one gallery. I personally have 11 move galleries at least twice that size. How long would it take to tag those?

Google crawls them.

What part of

Certainly not via the Galleries systems.

did you not understand? Especially when you quoted it. (IMG:[invalid] style_emoticons/default/dry.gif)

In-house solutions are best, especially because Google doesn't do all that the Galleries search can.

Plus, it's gonna take some expertise to use Google to search all E-H Galleries comments. And even then, there's no way to guarantee that Google's using a recently cached copy of a gallery.

Furthermore, if we suppose that such comments only contain valid tags (that is, tags that are also among the gallery's own "real" tags), then searching with Google will get you results absolutely no different than searching via the Galleries currently will. So, so long as we assume the comment "tags" are valid, your suggestion to use Google to search them is absolutely worthless. (And as soon as we assume they're invalid, then Google's usefulness becomes only about those tags that would be disallowed, not the otherwise valid ones.)

No generic third-party search engine is in any way a suitable substitute for a well-thought out and integrated system on the Galleries.

This post has been edited by PeopleDon'tDanceNoMore: Jun 5 2013, 08:04

especially because Google doesn't do all that the Galleries search can.

Technically Google gives more y'know... But obviously I can't say any more than that.

Technically Google gives more y'know... But obviously I can't say any more than that.

For anyone who knows what you're talking about, you're completely wrong. Google's still not going to give them any better results than they can get from a Galleries search - it may even give worse.

For anyone who doesn't know what you're talking about, you're right, but not in any useful way. Even if those people can find such "more", they'll still be lost in the end.

And in either case, by its nature, that means that Google can only ever cache very early versions of the "more". If those comment tags go up late (and they're likely to, I'd wager), then Google would never know of them, rendering your suggestion absolutely pointless for the "more".
(It looks like I could be wrong about this part, but it's hard to be sure. If so, I don't know how Google accomplishes it exactly.)

This post has been edited by PeopleDon'tDanceNoMore: Jun 5 2013, 09:12

And in either case, by its nature, that means that Google can only ever cache very early versions of the "more". If those comment tags go up late (and they're likely to, I'd wager), then Google would never know of them, rendering your suggestion absolutely pointless for the "more".
(It looks like I could be wrong about this part, but it's hard to be sure. If so, I don't know how Google accomplishes it exactly.)

Pretty true I guess.

For an example: say that a gallery is published at 13h, google crawls it at 13h 12mins and most of the tags and the comments are added at 13h 33mins. Google might take months before indexing the gallery again, and even after indexing google might keep the old one in the cache (the cache is used to generate the snippets).

Also, google still uses the bag of words model for search, so unless you do a very clever query these two are equivalent:
"chapter 1: tagA tagB; chapter 2: tagC, tagD" and "chapter 1: tagC, tagA; chapter 2: tagB, tagD"

This post has been edited by blue penguin: Jun 5 2013, 12:35

PDD's suggestion is good. One thing though, who should do the chapter splitting? The uploader (he might not do it), or maybe voting like with the rename system?
This would be very very, very useful for finding things in bigger tanks and magazines.

https://e-hentai.org/g/541034/1444043f65/
There's my random gallery, a mere 232 images right now. Now open a text document and use it to "tag" that gallery image by image and see how long that takes. Remember that this is just one gallery. I personally have 11 move galleries at least twice that size. How long would it take to tag those?

Tagging per image, I believe is really meant to be used for Dōjinshi, Manga, Western etc. that have no such well maintained online resource except for e-hentai. There a plethora of other sites for random images to search genres. So even though I don't have the proper system setup to do tagging the way I would want I did anyways.

https://e-hentai.org/g/541034/1444043f65/
Image# 1 Character: Horo Parody: Spice and Wolf Female: Animal Ears, Wolf Ears, Wolf Girl, Blush, Flat Chest, Brown Hair, Long Hair, Grin, Navel, Nude, Red Eyes, Deity Misc: Wheat  Artist: Tony Taka
Image# 2 Character: Belldandy Parody: Oh My Goddess Female: Blue Eyes, Angel, Goddess, Large Breasts, Earrings, Pubic Hair, Nude Misc: Draped Cloth
Image# 3 Character: Louise Francoise le Blanc de la Valliere Parody: Familiar of Zero Female: Blush, Cupping Breasts, Long Hair, Nude, Navel
Image# 4 Character: Yuffie Kisaragi Parody: Final Fantasy Male: Penis, Torso, Unidentifiable Female: Bottomless, Cum on Body, No Privates, Short Hair, Hand Job Misc: Sleeveless Turtleneck
Image# 5 Character: Sailor Mercury, Sailor Mars, Sailor Jupiter Parody: Sailor Moon Female: Blush, Long Hair, Short Hair, Blue Hair, Purple Hair, Brown Hair, Ponytail, Bondage, Nude Misc: Chains
Image# 6 Character: Mizuno Ami Parody: Sailor Moon Female: Blue Eyes, Blue Hair, Blush, Erect Nipples, Gym Uniform, Navel, Depantsed, Tanlines Misc: Locker Room, White Panties Artist: Mogudan

It took ten minutes for 6 images = 100 seconds per image @ 81 tags. Calculating this together with time taken it would take about 6.5 hour for your gallery alone.
However this also included time to look up photos on Google image search to find out who characters are what they're from etc. since I'm not sure who they are. I could only figure out 2 of the artists (I think?) If it were integrated with the website as I'm suggesting it would take far less time to input since I was using a spreadsheet. Please also remember that most individuals have there own tagging preferences and are always tagging what there niche genre(s) and who and what they know.
For image sets there are actually quite a few simple methods applied later on to help automate the process like autofills, a checksum or Google image search. A checksum processes would find these two or more identical files and populate tags from one to the others in the database in lets say a grey color = 1 total vote point on the images that are essentially duplicates. Using Google image search a automation script could be put together to find the same file in different sizes on the internet and use tags from other sites to help populate the lack of tags for the images. This is where you would probably use those sites for looking up that niche of random images.



During my tagging exercise I thought of some easier ways on how to use current existing system for a per image tagging gallery system. Trying to keep it simple use something along the lines of a double name space for the pages ex. Page001:Female:ahegao than users specifically see Female:ahegao in the main tagging part of the gallery where Page001 is non existent text until when Female:ahegao is clicked shows something like

Page001 Page122 Page123
Vote Up Vote Down Show Tagged Galleries Show Tag Definition Add New Tag

This would require a few changes on system which would be, first allow or figure out a double name space type system for new Page#: and second the namespace range from 0 or 1 to maximum number of allowed pages lets call it 999. So depending on how many images are in the gallery the namespace would go from Page1 to Page9, or Page01 to Page99, or Page001 to Page999, that's how Iv'e noticed the system numbers images.
Third Part is making the namespace link to just that Page# similar to how Show Tagged Galleries works.
Forth Part is easy enough adding the tagging text field "Enter new tags, seperated with a comma..."
the tricky part maybe having it add the Page#: namespace automatically for each page you are on when tagging.
Implementing a system like this you can slowly add it and all of other other necessary components on top of the existing system so it's not like a full system overhaul.
By no means is this series of ideas and steps the most elegant, best means or most complete solution to have a per image tagging system. Any constructive Ideas or Problems with this are free to fly...

This post has been edited by TJpregoRipper: Jun 5 2013, 14:52

PDD's suggestion is good. One thing though, who should do the chapter splitting? The uploader (he might not do it), or maybe voting like with the rename system?

With a voting system, you couldn't guarantee that some galleries wouldn't be split bigger or smaller than chapters (supposing you want the system to only be used for chapters). And if you had to redefine a bad division, dealing with previous tags on incorrect subgalleries would be a pain.

On the other hand, only letting privileged users split it up requires that there be enough users (who are motivated enough) doing this work. If you spread out the responsibility (and to an extent, I don't see why not), though, then it becomes less of a problem. (I say "I don't see why not" because, as I've mentioned before, I basically believe that anyone who might specifically ask for or accept that responsibility would be unlikely to abuse the privilege. Even if they did, that doesn't exactly make it much worse a situation than voting abuse, IMO. Nothing's perfect. =/ )

Before engaging in internet fisticuffs over usefulness and effort involved, shouldn't you at least make sure this is even feasible?
A change like that could put a lot of stress on the database
Just saiyan

With a voting system, you couldn't guarantee that some galleries wouldn't be split bigger or smaller than chapters (supposing you want the system to only be used for chapters). And if you had to redefine a bad division, dealing with previous tags on incorrect subgalleries would be a pain.

Yeah, probably not a good idea to leave it to voting.
I guess if some people volunteered to take care of the splitting it would be workable. Ideally uploaders would get the ability to do the splitting before publishing, too - that should probably reduce the resulting workload a lot, too.
Now as for what to use it for, chapters are one thing that's very easy to define. For other potential uses I'm not as sure.

Before engaging in internet fisticuffs over usefulness and effort involved, shouldn't you at least make sure this is even feasible?
A change like that could put a lot of stress on the database
Just saiyan

For the case of chapters, assuming an average of 18 chapters per magazine and 4 magazines uploaded per day, that comes to 72 additional galleries a day. Galleries uploaded per day seem to be about 125-200.
That's just a very rough estimation, but I'd say it would increase the number of galleries added to the system a day by about 1/2.
If it was implemented as a sub-gallery system, I'd hazard a guess there wouldn't be much additional overhead beyond the pure number of additional galleries in the DB.

Now, how much impact that would have is the question. If the sub-gallery system was only used in the search if you specifically activate it via a checkbox or something, that would reduce the additional strain by a lot in any case.

Tagging per image, I believe is really meant to be used for Dōjinshi, Manga, Western etc. that have no such well maintained online resource except for e-hentai. There a plethora of other sites for random images to search genres. So even though I don't have the proper system setup to do tagging the way I would want I did anyways.
It took ten minutes for 6 images = 100 seconds per image @ 81 tags. Calculating this together with time taken it would take about 6.5 hour for your gallery alone.

It doesn't matter how you would mean to us it, it would have to be open for and applied to all galleries. You've seen the absolutely ridiculous amount of time it would take to tag galleries. All your talk about how it could be sped up have a major problem, all tagging is done manually. Even if you came up with a script to rip tags from image sites to tag image sets, their entry system is different from ours and they allow a lot of tags that we don't. So even if you could reconcile the first problem, the second would result in a ban within a day or two.

It doesn't matter how you would mean to us it, it would have to be open for and applied to all galleries. You've seen the absolutely ridiculous amount of time it would take to tag galleries. All your talk about how it could be sped up have a major problem, all tagging is done manually. Even if you came up with a script to rip tags from image sites to tag image sets, their entry system is different from ours and they allow a lot of tags that we don't. So even if you could reconcile the first problem, the second would result in a ban within a day or two.

True the tags that are used in other sites are by far different other sites usually allow anything to be tagged. Tags in my test are including things we normally don't ever tag and would be similar in nature to other sites. If it's just character: parody: and 0-3 female: tags only than amount of time is greatly reduced. I found that almost (if not all) of the images you have on that particular gallery have duplicates running around the e-h galleries. So in my concept of the system tagging one would result in the other dozen or so duplicate images receiving that same tag that would at first appear grey and can be voted up or down like normal. Also not every single image needs a tag like in Dōjinshinji blanks, credits pages, images where it's setting up the scene, dialog, etc. Also tagging is a crowdsourcing area and it's been working so far. Why wouldn't it work even better with more ways to tag?
The best way to see what would really happen with tagging is to try out attaching the current tagging window on the bottom of the images and do a before and after # of tags comparison. Which wouldn't be tagging images just things in the gallery almost like it is now.

Also tagging is a crowdsourcing area and it's been working so far.

You haven't been tagging much if you honestly believe that.

With a voting system, you couldn't guarantee that some galleries wouldn't be split bigger or smaller than chapters (supposing you want the system to only be used for chapters). And if you had to redefine a bad division, dealing with previous tags on incorrect subgalleries would be a pain.

Before engaging in internet fisticuffs over usefulness and effort involved, shouldn't you at least make sure this is even feasible?

PDD's idea is pretty good. I guess the lesser effort would be to implement this split into subgalleries as simply new galeries. i.e. one big magazine (one gallery) is split into 12 chapters that are 12 galleries with the same uploader and timestamp. Although, a way of managing the tags would need to be invented.

a way of managing the tags would need to be invented.

Before anyone spends any more time trying to figure out a new system, ask Tenb. It's highly unlikely anything this big will implemented.

Before anyone spends any more time trying to figure out a new system, ask Tenb. It's highly unlikely anything this big will implemented.

It's honestly pretty simple to implement, if my guesses about how the current system is done are correct.

Before anyone spends any more time trying to figure out a new system, ask Tenb.

Why? If we decide we want to give our own time to it, then what the hell does it matter to you or anyone else?

Plus, as far as I'm concerned, "trying to figure [it] out" can only give Tenboro more information or ideas on how he might make it work, or on what we want. (And evidence suggests that Tenboro does take notice of this thread.)

If we decide we want to give our own time to it

"We"? (IMG:[invalid] style_emoticons/default/rolleyes.gif)

"We"? (IMG:[invalid] style_emoticons/default/rolleyes.gif)

Who are you supposing that refers to?

I simply meant anyone (which includes myself) who "spends any more time trying to figure out a new system". I.e., those of us deciding to engage in that discussion. What does it matter if we want to have that discussion?

This post has been edited by PeopleDon'tDanceNoMore: Jun 7 2013, 08:06

A lot of these ideas are just someones pipe dreams until otherwise said but it's still a dream. Things that may or most likely not going to happen, will be and should be discussed and detailed. I do love the fact this site does have a suggestion box area. I've even thought of adding more fancy things like "Adding polls to suggestion box", to see what users might want more to improve the website. Heh however, I don't think with as much maintenance that this site regularly needs someone as busy as Tenboro can add tons of awesome features, and as it is the system work pretty well. But we can still dream discuss and detail...

On a side note to that I do wonder if anyone is/has freely offered website programming services to Tenboro to help get things either added or fixed. Obviously it is an area of large responsibility and trust.

This post has been edited by TJpregoRipper: Jun 7 2013, 12:06