[Prototype] User Report Ver 6.3

Jarlath said:

I personally consider tagme on a post with more than 10 tags an error in tagging, especially since 90% of what it covers can be handled by various request tags. For posts with less than 10 tags, tagme is just lazy as hell and is something I think should be weighted at a -10 for tagging report purposes, if not more. I've seen users throw tagme into a post to skip tagging them in a few occasions.

Well, I already give those users a neutral feedback. I kinda like this tag and feel "responsible" for the correct usage of it :3.

@provence @kuuderes_shadow

Sorry, I forgot to include it in the tables update... the removes column disregards the following tag removals: tagme, commentary, check_commentary, translated, check_translation, partially_translated, check_my_note, check_pixiv_source, *_request. I'll update the version notes with that info now.

As I mentioned, the reason why the removes column is kept separate from the tag errors, is because I saw a certain amount of subjective changes such as hair color changes for instance. But there were also a lot of genuine errors, for example the wrong character tag, or tagging headband when it was actually a hairband.

As for Jarlath's recommendation, I can include the adding of tagme to posts with 10 or less tags as a Tag Error, or I could make them a separate column (although the table is quite wide as it already is... :)

Edit:

Also, Tag Errors are currently mutually exclusive from the other tag categories, whereas Removes are not. So in forum #118765, kuuderes_shadow correctly subtracted the Removes category, but incorrectly subtracted the Tag Errors.

For that, I can go either way. I can leave it as it is, I can make it so that both are included in the other tag categories, or I can make them both mutually exclusive from the other tag categories.

Thoughts?

Also, after thinking about it for a bit, Tagme Errors would probably be better as a separate column, since otherwise it'd be mixing with and polluting the Tag Errors column, as it would be impossible to tell which one actually caused the error just from the data.

I decided to create a new forum post since the last one was busy as is... :p

kuuderes_shadow said:

It might be interesting to have a top (and bottom) tags/upload list for everyone - if possible with errors and removes subtracted from the list. For obvious reasons there would need to be a threshold for the number of uploads to stop both lists being flooded with people with 1 or 2 uploads. 100 is more than enough for this. 50 would probably be fine, too.

The one issue with this is that the number of uploads should factor into the weight of the rating.

Example:

User A uploads 500 posts with only 10 total tags per upload
User B uploads 10 posts with only 10 total tags per upload

They are both equal when you reduce them to a per upload metric, but one is clearly more damaging and causing more work for other tag gardeners. How do you represent that as a legitimate and quantifiable metric? I'm asking because I don't know... (Math was only my minor, and not my major...)

Don't know what you exactly, but if someone really has 500 uploads and has very few tags added, the existing tables are sufficient. You of course has to look what they upload. I nearly never get over a gentag of 15 when I upload Pokémon fanart for example. So yeah, numbers and systems are good, but one should never forget that in order to work with these tables, you have to do work manually, too.

The only difference to the rankings made by not subtracting the errors is that DeusExCalamitus becomes 0.05 tags per post 'better' than RazingK, rather than 0.06 'worse'. Not worth reposting over that.

As for the second post - I can think of various ways of doing it, but honestly, I don't think it's justified as:
- Most of the methods would involve an arbitrary method for selecting how much influence the tag count has on the figure, and would create very different results depending on what method was chosen. In some cases, you'd end up with a figure that doesn't make any sense to the casual observer as well.
- Someone who consistently uploads 10 posts a month with 10 tags per upload is just as much in need of improving their tagging as someone who does 500 posts a month with 10 tags each.
- So long as the requirements for inclusion are high enough for margins of error to be immaterial, a decent method shouldn't have an overly large impact on the figures anyway.

Just including the number of uploads on the table would be sufficient for anyone who deems that to be important.

Why the edit? What changed :<?

New version... you can check the version info in the first post, but basically, I added correlation between wiki page changes made through the artist interface. This added an extra column to the artist table. For the wiki pages table, it disregards any changes made through the artist interface. This updated the numbers and rankings for both tables.

Also, I hadn't planned on creating a new forum post (like this) with every change to the first post, but I guess I could...

What table did you update? They are all from yesterday or two days (at least in CET^^), but I guess the artist table changed then^^?

Artist and Wiki Page tables were updated.

Artist table with a new column "Wiki" on edits to the wiki through artist interface.

Wiki Page table without edits through artist interface.

I'm kinda scratching my head why I have an average score of -3 in the comment section. Well, I get that my comments won't get much score (and they also shouldn't in my eyes xD), but is there a quick way to find which comments were downvoted so drastically. Probably because I argued and discussed a lot on flagged posts and defended the flag :>.

Provence said:

I'm kinda scratching my head why I have an average score of -3 in the comment section. Well, I get that my comments won't get much score (and they also shouldn't in my eyes xD), but is there a quick way to find which comments were downvoted so drastically. Probably because I argued and discussed a lot on flagged posts and defended the flag :>.

Probably not. But here they are anyway:

RaisingK said:

Probably not. But here they are anyway:

Show

post #2459629 (Sat Aug 20 00:43:29 EDT 2016) = -5
post #2457014 (Wed Aug 17 10:17:25 EDT 2016) = -1
post #2457014 (Wed Aug 17 10:08:37 EDT 2016) = -1
post #2455834 (Tue Aug 16 19:00:19 EDT 2016) = -3
post #2365123 (Tue Aug 16 07:06:25 EDT 2016) = -1
post #2455834 (Tue Aug 16 05:13:11 EDT 2016) = -1
post #2455834 (Tue Aug 16 04:36:44 EDT 2016) = -5
post #2454067 (Sun Aug 14 22:26:48 EDT 2016) = -4
post #2454151 (Sun Aug 14 11:20:19 EDT 2016) = -1
post #2452910 (Sat Aug 13 17:08:48 EDT 2016) = -1
post #2451619 (Thu Aug 11 18:08:59 EDT 2016) = -1
post #2441849 (Tue Aug 02 10:14:13 EDT 2016) = -1
post #2441484 (Tue Aug 02 02:39:31 EDT 2016) = -6
post #2439501 (Sun Jul 31 15:55:40 EDT 2016) = -1
post #2435635 (Wed Jul 27 14:29:29 EDT 2016) = -1
post #2435571 (Wed Jul 27 10:28:43 EDT 2016) = -1
post #2431987 (Sat Jul 23 18:38:52 EDT 2016) = -1
post #2431987 (Sat Jul 23 18:27:59 EDT 2016) = -1
post #2423174 (Fri Jul 22 04:17:43 EDT 2016) = -1
post #1378109 (Thu Jul 14 07:23:50 EDT 2016) = -1
post #246578 (Thu Jul 14 07:20:16 EDT 2016) = -1
post #2398721 (Thu Jul 14 07:12:21 EDT 2016) = -1
post #2421577 (Wed Jul 13 12:52:19 EDT 2016) = -1
post #2420594 (Tue Jul 12 14:47:51 EDT 2016) = -1
post #2420594 (Tue Jul 12 14:27:22 EDT 2016) = -1
post #2416394 (Fri Jul 08 10:52:37 EDT 2016) = -1
post #2409334 (Fri Jul 01 11:44:24 EDT 2016) = -1
post #2407369 (Wed Jun 29 19:43:51 EDT 2016) = -1
post #2403230 (Sun Jun 26 13:02:32 EDT 2016) = -1
post #2401606 (Thu Jun 23 15:12:59 EDT 2016) = -1
post #2401606 (Thu Jun 23 14:49:49 EDT 2016) = -1
post #2401606 (Thu Jun 23 14:42:24 EDT 2016) = -1
post #2399353 (Tue Jun 21 09:51:42 EDT 2016) = -1
post #2391155 (Sun Jun 12 18:45:26 EDT 2016) = -5
post #2390997 (Sun Jun 12 15:01:54 EDT 2016) = -1
post #2387889 (Fri Jun 10 04:37:30 EDT 2016) = -4
post #2387662 (Thu Jun 09 16:42:36 EDT 2016) = -2
post #2386545 (Wed Jun 08 12:11:32 EDT 2016) = -1
post #2384696 (Tue Jun 07 02:47:20 EDT 2016) = -1
post #2383879 (Mon Jun 06 06:05:25 EDT 2016) = -1
post #2383261 (Sun Jun 05 18:33:16 EDT 2016) = -1
post #2383260 (Sun Jun 05 18:28:04 EDT 2016) = -1
post #2383261 (Sun Jun 05 18:26:20 EDT 2016) = -1
post #2383261 (Sun Jun 05 14:52:28 EDT 2016) = -1
post #2383261 (Sun Jun 05 14:38:32 EDT 2016) = -1
post #2383270 (Sun Jun 05 14:30:50 EDT 2016) = -1
post #2383261 (Sun Jun 05 14:17:24 EDT 2016) = -1
post #2373172 (Tue May 31 07:24:07 EDT 2016) = -1
post #2373343 (Fri May 27 14:52:28 EDT 2016) = -2
post #2364444 (Wed May 18 15:52:46 EDT 2016) = -1
post #2316331 (Sun May 15 12:08:14 EDT 2016) = -1
post #2360899 (Sun May 15 01:56:30 EDT 2016) = -1
post #2348459 (Tue May 10 11:36:07 EDT 2016) = -1
post #2355550 (Mon May 09 09:39:51 EDT 2016) = -2
post #2355550 (Mon May 09 09:29:26 EDT 2016) = -4
post #2355550 (Mon May 09 09:10:25 EDT 2016) = -3
post #2354126 (Sat May 07 21:49:19 EDT 2016) = -1
post #2354126 (Sat May 07 20:12:28 EDT 2016) = -1
post #2349462 (Tue May 03 09:38:35 EDT 2016) = -1
post #2348785 (Mon May 02 15:15:50 EDT 2016) = -1
post #2346009 (Mon May 02 07:36:12 EDT 2016) = -1
post #2345544 (Fri Apr 29 21:50:55 EDT 2016) = -1
post #2344631 (Fri Apr 29 06:33:26 EDT 2016) = -1
post #2344090 (Thu Apr 28 12:40:57 EDT 2016) = -3
post #2342275 (Tue Apr 26 16:16:53 EDT 2016) = -4
post #2339130 (Sun Apr 24 22:09:27 EDT 2016) = -3
post #2337137 (Thu Apr 21 16:00:18 EDT 2016) = -2
post #2332322 (Sat Apr 16 11:17:22 EDT 2016) = -1
post #2330250 (Thu Apr 14 08:47:33 EDT 2016) = -1
post #2306719 (Sat Apr 09 15:27:30 EDT 2016) = -1
post #2325365 (Fri Apr 08 18:08:14 EDT 2016) = -2
post #2324016 (Thu Apr 07 02:53:34 EDT 2016) = -1
post #2300063 (Tue Apr 05 10:59:57 EDT 2016) = -1
post #2321857 (Mon Apr 04 19:01:56 EDT 2016) = -2
post #2318263 (Mon Apr 04 06:11:57 EDT 2016) = -1
post #2318263 (Sun Apr 03 12:23:29 EDT 2016) = -1
post #2318263 (Thu Mar 31 14:25:38 EDT 2016) = -1
post #196630 (Tue Mar 29 19:02:39 EDT 2016) = -1
post #2314737 (Sun Mar 27 14:10:11 EDT 2016) = -2
post #2299406 (Wed Mar 09 18:04:23 EST 2016) = -1
post #2299382 (Wed Mar 09 17:31:18 EST 2016) = -4
post #2295947 (Mon Mar 07 09:35:19 EST 2016) = -1
post #2291630 (Tue Mar 01 17:38:54 EST 2016) = -1
post #2276543 (Mon Feb 15 19:45:57 EST 2016) = -1
post #2271146 (Wed Feb 10 19:21:07 EST 2016) = -1
post #2260524 (Sat Jan 30 11:19:40 EST 2016) = -1
post #2208092 (Sat Dec 12 03:31:57 EST 2015) = -1

So it's most often harmles like asking things which is apparantly not well received^^. Well, I guess I'll stop that then if I should have a question about a character or if I disagree with something.

RaisingK said:

Probably not. But here they are anyway:

Did you code that on the fly, or did you just have that handy...? :)

BrokenEagle98 said:

Did you code that on the fly, or did you just have that handy...? :)

Well about that: Would that be possible to implement?
I mean for example:

Provence said:

Well about that: Would that be possible to implement?
I mean for example:

Are you asking to have two columns added to the comments table, namely the number of positive comments and negative comments? Everything else you need is already there...

Average score = cumulative score/ total comments
Deleted comments is already a column

BrokenEagle98 said:

Are you asking me to add two columns to the comments table, namely the number of positive comments and negative comments? Everything else you need is already there...

Yes. That was more or less a final version. So the comment section would look like this then:
Rank Username Total Updates Positive Negative Cumulative_Score Average_Score Deletes Bumps

By positive and negative the total amount of neg/pos comments.

Ok... I'll probably leave average score out though, since that's just basically one column divided by the other....

@BrokenEagle98: Not RaisingK, but I use jq for stuff like this. Makes quick and dirty JSON processing a snap:

$ curl -s 'http://danbooru.donmai.us/comments.json?group_by=comment&search%5Bcreator_id%5D=460797&limit=1000' | jq 'sort_by(.score) | map({url: "http://danbooru.donmai.us/posts/\(.post_id)", score, body}) | .[0:5]'     
[
  {
    "url": "http://danbooru.donmai.us/posts/2441484",
    "score": -6,
    "body": "Now where can this image be found on his personal website? I don't really see a link to it (like the source directs the the image.)\r\nAlso...shouldn't ricegnat then be aliased to his real name which is Eric Tang I think?"
  },
  {
    "url": "http://danbooru.donmai.us/posts/2459629",
    "score": -5,
    "body": "A 22 score over a bit of n hour? Makes me feel sad since I wasn't stalking pixiv although this here came up multiple minutes after the pixiv upload (probably with a md5 change, though :3)"
  },
  {
    "url": "http://danbooru.donmai.us/posts/2455834",
    "score": -5,
    "body": "[quote]\r\n\r\nRandeel said:\r\n\r\nThat's just the style of the artist, Provence.\r\n\r\n[/quote]\r\n\r\nWell, I'd gladly flag those other posts, too....maybe^^. But if it gets approved, I think this \"style\" is accepted here ;)."
  },
  {
    "url": "http://danbooru.donmai.us/posts/2391155",
    "score": -5,
    "body": "(Nearly) Perfect as always^^. "
  },
  {
    "url": "http://danbooru.donmai.us/posts/2454067",
    "score": -4,
    "body": "Oh really? \r\nIt's not really the pixelated thing (although it doesn't look good in my eyes). This is weird yes, but also the anatomy is not that great. The left arm's elbow looks way to far away from here body. Since the arm goes up in a 90 degree angle, her elbow must be located on the same line you draw, when you want to connect the left knee and the left hand...and that's not good.\r\n\r\nThat's why post #2435813 wasn't flagged. It doesn't show this flaw, although is more or less the exact same pose. "
  }
]

If other people want to play with this:

• Copy your comments from here (replace the ID with your own; limit can go up to 1000): http://danbooru.donmai.us/comments.json?group_by=comment&search%5Bcreator_id%5D=460797&limit=100
• Paste that into https://jqplay.org/.
• As the filter, use this to sort your comments by score:

sort_by(.score) | map({url: "http://danbooru.donmai.us/posts/\(.post_id)", score, body})

• Percentage of comments that are negative:

(map(select(.score < 0)) | length) / length

• Cumulative score:

[.[] | .score] | add

• Average comment score:

[.[] | .score] | add / length

Aw snap... that's pretty cool...

I'll have to check it out...

@Provence @kuuderes_shadow

Updated user report:

Ver 4.2 -> Ver 4.3

• Added top/bottom 25 upload tagger tables <kuuderes shadow: forum #118765>
• Added count for positive/negative comments in comment table <Provence: forum #118825>