Topic: Tagging text-based tags. (Tag suggestions/improvements appreciated!)

Posted under Tag/Wiki Projects and Questions

Pup

Privileged

I figured I've done threads for almost everything else I'm tagging, so I might as well make one for this.

I'd like to start by saying/asking that if I've missed any that I could tag, please let me know. I'd greatly appreciate it.

I'm currently using text recognition to tag:
text
english_text
url
pet_praise
apology
grawlixes
dirty_talk
profanity
?!
greeting
holiday_message

For english_text I've got a small dictionary of common words that are over four characters long. Any less and Chinese/Japanese characters cause mis-tagging.

For apology I'm tagging posts that include "apologies", or "sorry" without "but". Does that seem ok, or should I change it? Or are there others I should add?

For dirty_talk I looked over the tag and picked a few phrases, such as "fill me with your" and including "cum/seed". I've got one or two, but any you can think of would be helpful.

For greeting I've mostly stuck with "hello", I originally had "hey" but it could be more of a shout of surprise. "Hi" seemed good until it tagged any skyscraper with it, as the concrete between windows make "H" and "I" shapes, like with post #1875087. "Sup", I thought was a bit vague, " 'sup" could be better. I'm also tagging "gday" and "g'day".

And for holiday_message I've tagged most of them, but should I tag mother's/father's day? They're specific days, but not really holidays. Also, mother's_day implies holidays, but I'm still unsure if "happy mother's day" is really a holiday_message.

Then these are some I'm not so sure how to tag:
question
threat
wall_of_text

For question, could I tag anything with a question mark? I felt like that was a bit obvious, but wasn't sure whether to say it also had to include, "what/why/when/who/how/are you", or if the question mark was enough.

For threat, I can't think of how to tag it. I thought of different phrases, like dirty_talk, but often they could be used in other contexts. Something like "do x or I'll y", seems ok, but I'm still uncertain on it.

And lastly, wall_of_text, at first I thought it'd be easy, count the words in the current wall_of_text posts, then apply that to others. But then it's surprising how many comics would get mis-tagged, that definitely aren't walls of text. So I feel there isn't an easy way to tag this, as it's more on how much the text fills the image, rather than just the character count.

Again, any ideas for alterations, improvements or more things I could tag would be greatly appreciated!

Edit:
I included "(Tag suggestions/improvements appreciated!)" in the title as I didn't want people thinking I was asking how to go about automating the tagging, rather than looking for suggestions, though I'm not keen on the wording.

Updated

Pupslut said:

For apology I'm tagging posts that include "apologies", or "sorry" without "but". Does that seem ok, or should I change it? Or are there others I should add?

what about sarcasm? like 'Gee I'm soo sorry that you're an asshole' isn't what I'd call an apology

For dirty_talk I looked over the tag and picked a few phrases, such as "fill me with your" and including "cum/seed". I've got one or two, but any you can think of would be helpful.

What if it's not really in a sexual context? Like a character giving another one tips for how to spice up their sex life? I don't know if this would count or not.

For greeting I've mostly stuck with "hello", I originally had "hey" but it could be more of a shout of surprise. "Hi" seemed good until it tagged any skyscraper with it, as the concrete between windows make "H" and "I" shapes, like with post #1875087. "Sup", I thought was a bit vague, " 'sup" could be better. I'm also tagging "gday" and "g'day".

just be careful so it doesn't mean stuff in other languages, or in a different context, like 'hi res' for example.

Updated by anonymous

Pup

Privileged

MyNameIsOver20charac said:
what about sarcasm? like 'Gee I'm soo sorry that you're an asshole' isn't what I'd call an apology

I never really liked tagging that one, it didn't mis-tag, but I kept checking it anyway. There's a few ways around it, but I'd rather not tag it at all.

MyNameIsOver20charac said:
What if it's not really in a sexual context? Like a character giving another one tips for how to spice up their sex life? I don't know if this would count or not.

I'll have to check the other lines I've included, but I think this one should be ok.

MyNameIsOver20charac said:
just be careful so it doesn't mean stuff in other languages, or in a different context, like 'hi res' for example.

Oh, I am, that's why the dictionary is only words that are 4 characters or longer. Chinese/Japanese often get recognised at 2-3 length English words, also it's set so it won't tag anything with english_text if it's already got *_text already tagged.

"Hi res" would still be English, though like I said, it's avoiding the smaller words, so that should cover those types as well.

Thanks for pointing those out, I appreciate it, and if you think of any tags I missed, that could be tagged, then let me know.

Since I have to download each image to run text recognition on it I've set it to only check images once, which is mostly the reason I wanted to make this thread, so I don't miss loads then have to go back over the images I've already checked. Though I'll have to do that anyway when I have another go at tackling cool_colors.

Updated by anonymous

Pupslut said:
I never really liked tagging that one, it didn't mis-tag, but I kept checking it anyway. There's a few ways around it, but I'd rather not tag it at all.

I meant that I'm not sure if it would qualify for the apology tag

Also kudos on making this bot! I've seen lots off people discuss if it would be possible to make one and what it could tag in that case, but you're the first one I've seen actually making one. Great work!

Updated by anonymous

Pup

Privileged

MyNameIsOver20charac said:
I meant that I'm not sure if it would qualify for the apology tag

You mean that you meant sarcasm wouldn't count as an apology?

I meant that I didn't like tagging the apology tag, not sarcasm, as I felt there was several times someone could say "sorry" but not mean it. Such as, "I'm sorry but.." So I worked around that, but didn't think about sarcasm, so it's a good job you mentioned it. I was saying that I'd prefer not to tag apology, as it was too hard to tell the context of it.

MyNameIsOver20charac said:
Also kudos on making this bot! I've seen lots off people discuss if it would be possible to make one and what it could tag in that case, but you're the first one I've seen actually making one. Great work!

Thank you! It really means a lot! I'm definitely putting a lot of effort into it so I'm glad someone thinks it's worth it. :)
There's been a few others, such as TagBot, but I'm hoping to include more tags than they did. I know they didn't include text based tags, so that's already a plus.

It can be frustrating at times, such as with the black_and_white/greyscale tags, but I think it'll be worth it in the long run.

(Shameless poke for the thread talking about those tags, forum #272717, in case anyone else reading this has some input. For now it's more about what counts as black_and_white, and greyscale, without thinking about spot_color for now. I/we can think about that after people agree on the monochrome aspect.)

While I'm thinking about it, would you consider "happy father's day" to be holiday_message? Father's day implies holidays but I've never really thought of it that way.

And thanks again, both for the kind words and for posting on most of these threads to correct me and stop me mis-tagging, I really appreciate it!

Updated by anonymous

  • 1