Problem:
When I search for, an example, "female renamon" (searching for all female renamon pictures, not just solo), I will get results that include male renamon on another female character.
So, I was thinking, how could this problem be solved, and I've come to several solutions.
1. Some tags could be linked to a gender.
This would work with species or character names. So basically, whenever a picture would have a character name or a species, the user could assign a gender to it. Then I could search for "Rainbow_Dash-Male" to hide all the female versions of the said pony.
But it wouldn't work in cases like "female nude" where you would still get to see a dressed female character, with a nude male next to her.
2. Tags could be split per character
The tags would be split into Character tags and General. In the character tags, there would be a list of characters, and each character would have its own tags. This way I could search for "(female nude)" to say that both tags must appear on the same character.
Looks good! But, still not perfect. What if I want to search for "male female herm threesome" where male is having an intercourse with the herm?
3. Tags split per character + interaction type
We're able to define characters, and interactions between them.
This would be great because now we can search for specific character and the way they interact. Plus, the system could use this logic to deduce a lot more tags on its own.
For example, you have FFM picture (male on female, female licking another female). You select that image has 3 characters, female1+female2+male1. Then you select interactions female1-female2 + female2-male1. And voila! The system itself defines female1=lesbian, female2=bi, male1=straight.
Or, you have pokemon-trainer picture and you define the interaction as sexual. Then system would then infer the tag "pokephilia".
There is so many possibilities.
As for searching, searching for posts that include any male character interacting with female pony: "(male)-(female pony)".
Expanding on the concept we could go as far as:
4. Object Modelling:
-Any number of definable objects with object properties
-Any number of definable interactions with interaction properties
..but I dunno if I want to go that far.
Edit:
End notes:
The way current tagging works is like painting a picture on a single surface. It's like trying to paint a full-color piece without sketching it first. Yeah, some of the best artists are able to do that, but what about the rest? The rest of normal artists can't do that and they use more structured process. Sketching, inking, coloring, shading etc, use several layers and what not. And this is how tagging should work as well. It should be an organized process which anyone can easily use to produce a perfect representation of an image. But currently only taggers with years of experience can do that.
Now users will say "New system would make tagging slower!". Not necesarrily. New system could infer a lot more from the characters and their interactions than it currently can from a simple array of tags, which would mean less work.
Besides this, tagging is as fast as you make it to be. The longer you spend tagging, the better the image is described. The whole purpose of tagging is to try to accurately describe an image so that it can be easily found, right? And if you can quite simply descibe an image better, wouldn't you want to spend a few seconds more to do that?
Updated by Fluttershy