Also, is it feasible to have a Ruby interpretation of what BrokenEagle did to get those stats and incorporate it into the Danbooru website since it runs on Ruby?
That's the idea, although I don't know Ruby that well myself to give a good opinion on how hard/easy it would be. The above is just an interim thing until/unless a server-side Danbooru report ever gets implemented.
Well, of course this is a technical question. So I'd by happy to hear what @Type-kun or @Toks have to say about this in the technical way to implement this.
So if I'm interpreting this correctly, the promotion lists are meant to encourage people to do more of whatever they're doing so they get a promotion? I suppose that could work, though I suspect most people who are heading for a promotion will continue to do whatever they're doing and get said promotion, while those who aren't even close won't care enough to check the lists. If it encourages a little more positive activity, though, then no harm in implementing it. It may always serve as a way of curating information on board member stats, of course.
Regarding 9. The top N tags added or removed can be calculated per user as well. Is that something that should be shown, and if so, what should N be?
It would have to be at least 20 if it's to give any useful information, I reckon, as many people would probably be able to guess their top few (eg. my top added would probably include greyscale, solo, solo focus, quaver, comic, neck ribbon, red ribbon, ribbon, multiple views...)
These reports have existed for years actually, they're on a page called the mod dashboard. I think its limited to mods because of performance reasons. It's configurable with options like how far back to count changes and what levels of users to include.
It's missing pool changes and artist commentary changes though. Probably not hard to add but performance should be tested.
These reports have existed for years actually, they're on a page called the mod dashboard. I think its limited to mods because of performance reasons. It's configurable with options like how far back to count changes and what levels of users to include.
Could a report that is only updated like once a week (or month) be made available to regular users?
Could a report that is only updated like once a week (or month) be made available to regular users?
Exactly my thoughts, since users who are good translators can better evaluate the work of users below Builder rank and give report to moderators if they should get promoted or give feedback, poitive, neutral or negative. And a larger amount of users who can see that is always better if the performance can keep up^^.
Awesome, that sounds great! Before starting anything server-side though, I'd like for us to spend a few cycles in spiral development going over the look and feel like we have been, so that the Danbooru coders can tailor what you have to what we want, as long as it's not too difficult to implement.
I'll be throwing up some new versions of the reports based upon the feedback received so far later this afternoon/evening...
So I guess then there are no really obstacles in our way to implement these reports. Still, one last question that was a bit unnoticed: forum #118378 If we make this public, then we should add "Report" to this bar so it can easily be accessible from everywhere on Danbooru.
For the remaining categories, I've no clear opinion, so maybe calling the top users on each list here, so that they can give a good benchmark if they approve of this. And as I said: The mark should be different for every category. [...] Artist Changes: @RaisingK@tapnek
tapnek said:
I just do a lot of artist changes because I find myself uploading art from a lot of new artists and finding new information on artists frequently. Having good Google-fu helps. As for a benchmark, I'm the wrong guy to ask for that.
A lot of my edits are automated, so I'm even worse for judging benchmarks.
First off, you'll note the addition of two new categories: forum posts and comments. Forum posts gives an indicator on those very active on the forums, either answering questions, asking questions, or otherwise helping with the running of Danbooru. Comments gives us an indicator of those very active at participating in the more social aspects of this site, an important aspect IMO.
Second, I changed the cutoffs, but I purposefully left some of them low to give a good idea on what different cutoffs would do to the size of the tables. The cutoffs are: {pool:25, note:200, artist_commentary:25, artist:25, wiki_page:25, post:500, comment:100, forum_post:25}
Third, I combined both tables into one table for each category. I highlight the Platinum- users by italicizing their names.
Fourth, I didn't highlight it last time, but the hyperlinks in the first column lead to each user's page.
Fifth, I added hyperlinks to the second column, which lead to the search pages for each user/category combination.
Note: Some of the data changed for the Post Changes table (forum #118371). I was unfortunately working with incomplete data for the prior showing. The current table in this post contains the complete data.
Take a good look at the overall presentation of the tables. Also compare the size for each category. Please comment if you have any ideas to make it look better or be more useful. Any feedback will help the Danbooru coders out for when the server version gets implemented. Thanks.
Great work @BrokenEagle98. Looks exactly how I wanted it to look when I created this topic. Now thev last thing is to implement this and let it actualize itself for every month^^. And since it should be easy to access, this Reports should be shown somewhere. Like I said, that bar would be good if it is possible to edit.
Added a lot more data to each of the categories with versions (i.e. not comments or forum posts). Also adjusted some of the cutoff levels for some of the categories.
Certain stats are similar between all version categories.
Total: The total number of versions created by each user.
Create/Upload: The total number of new instances by each user.
Other: The total number of versions created with no changes, e.g. pools do not track description changes.
Besides the above, each category has unique stats for the various kinds of edits. Multiple types of edits can exist for a single single version change, e.g. adding and removing tags on a single edit. So basically, sum(edit events) != sizeofunion(edit events).
Also worth noting, Create events, Other events and union(edit events) are mutually exclusive. Therefore, to figure out the size of union(edit events), aka Post/Pool/Note/etc. Editing, subtract the Create and Other events from the Total events.
Example: Post Edits = Total Post Changes (third column) - Upload/Creates (fourth column) - Other Changes (last column)
Below the tables are the implementation details client-side using psuedocode to help out the Danbooru coders. There's probably an easier way to do things server-side, but it's being included just in case.
Data
Post Changes - Updated at Sat Aug 20 02:29:17Z 2016
Uploads: (length of unchanged_tags == 0) and (length of added_tags > 0)
Tag adds: (length of unchanged_tags > 0) and (length of added_tags > 0) and (count of rating/source/parent metatags) < (count of added_tags)
Tag removes: (length of unchanged_tags > 0) and (length of removed_tags > 0) and (count of rating/source/parent metatags) < (count of removed_tags)
Rating: ("rating:" in added_tags) and ("rating:" in removed_tags)
Parent: ("parent:" in added_tags) or ("parent:" in removed_tags)
Source: ("source:" in added_tags) or ("source:" in removed_tags)
Other: (length of added_tags == 0) and (length of removed_tags == 0)
could also be done by checking if none of the conditions above are true
Pools
Total: total versions/user
Creates: no prior version
Order: remove all posts not in intersection(priorver,currver); compare order
Adds: currver has posts not in intersection(priorver,currver)
Removes: priorver has posts not in intersection(priorver,currver)
Other: currver == priorver
could also be done by checking if none of the conditions above are true
Notes
Total: total versions/user
Creates: no prior version
Edits: body changed
Moves: x,y changes
Resizes: width,height changes
Deletes: isactive "True" -> "False"
Undeletes: isactive "True" -> "False"
Other: no changes, i.e. none of the conditions above are true
Commentary
Total: total versions/user
Creates: no prior version
Orig Title: Original Title changed
Orig Descr: Original Body changed
Trans Title: Translated Title changed
Trans Descr: Translated Body changed
Other: no changes, i.e. none of the conditions above are true
Artist
Total: total versions/user
Creates: no prior version
Name: name changed
Other Name: other names changed
Url: url string changed
Group: group name changed
Deletes: isactive "True" -> "False"
Undeletes: isactive "True" -> "False"
Other: no changes, i.e. none of the conditions above are true
Wiki Page
Total: total versions/user
Creates: no prior version
Title: title changed
Other Name: other names changed
Body Edits: body changed
Other: no changes, i.e. none of the conditions above are true
Final Thoughts
Like before, any comments on the look or feel of the tables above will be helpful. This includes column order, column name, table order, cutoff values, italicising vs underlining, columns to add (if possible) or remove, etc...
Edit:
3.0->3.1: Changed usage of italics for Platinum- group to underlines. (ref forum #118448)