Wikidata:Requests for permissions/Bot/MsynBot 7
The following discussion is closed. Please do not modify it. Subsequent comments should be made in a new section. A summary of the conclusions reached follows.
- Approved --Lymantria (talk) 07:26, 8 May 2021 (UTC)[reply]
MsynBot 7 edit
MsynBot (talk • contribs • new items • new lexemes • SUL • Block log • User rights log • User rights • xtools)
Operator: MisterSynergy (talk • contribs • logs)
Task/s: mark unpatrolled, but meanwhile overwritten changes by IP editors and newcomers as patrolled
Code: not yet complete; will be Python/pywikibot and based on https://hub.paws.wmcloud.org/user/MisterSynergy/notebooks/misc/2020%2010%20unpatrolled%20changes/patrol.ipynb
Function details: Quite a lot of unpatrolled edits are being overwritten or removed without being marked as patrolled. This applies to vandalism as well as good edits and leads to an unnecessary patrol workload for recent changes patrollers.
The bot would query unpatrolled changes from the database and then check whether the corresponding information is still available in the item using pywikibot. If not, it marks the unpatrolled revision as patrolled, since no experienced editor needs to review the revision any longer. It does not matter whether the unpatrolled revision was an act of vandalism or not.
I am currently able to patrol:
- reverted changes (based on the "mw-reverted" tag)
- sitelink additions and removals
- label additions, modifications, and removals
- description additions, modifications, and removals
Furthermore I am planning to go over alias and claim modifications as well, but particularly the latter can be a bit more challenging. I will, of course, only add them when I am highly confident about the outcome.
I expect to patrol a five-figure amount of revisions per month, or 10–30% of all unpatrolled changes. The script would probably run once a day from Toolforge. I have been running the testing script that I am deriving the bot from (see PAWS link above) under my regular account for a couple of days now, so you can see it in action in the patrol log. —MisterSynergy (talk) 17:27, 7 May 2021 (UTC)[reply]
- I'm not sure I understand the value. Do people actually use the patrol feature here? Or do they not because of this? I don't have a problem with this just confused. BrokenSegue (talk) 17:44, 7 May 2021 (UTC)[reply]
- We do not use it as much as needed, and there are several reasons for this. Poor tooling, difficulty of the job (for inexperienced patrollers), low efficiency—and the fact that way too often one reviews changes that have actually been changed again, which is plain unnecessary. I am targeting to alleviate the latter problem here.
As I have been patrolling probably way more than anyone else in the past months—a good part of that was already automated—I think I do have a feeling for the situation and see potential for improvements here. I have now filed a bot task because I want to fully automate the execution of such a script. —MisterSynergy (talk) 18:07, 7 May 2021 (UTC)[reply]
- We do not use it as much as needed, and there are several reasons for this. Poor tooling, difficulty of the job (for inexperienced patrollers), low efficiency—and the fact that way too often one reviews changes that have actually been changed again, which is plain unnecessary. I am targeting to alleviate the latter problem here.
- Strong support diminishing the amount of edits to be patrolled by humans is a big help. --Epìdosis 19:39, 7 May 2021 (UTC)[reply]
- Support Excellent, this will really help! I just marked patrolled a bunch of reverted revisions by an IP user earlier today, this would have made that unnecessary! ArthurPSmith (talk) 21:01, 7 May 2021 (UTC)[reply]
- Support This will be really helpful for patrolling recent changes. --Ameisenigel (talk) 06:25, 8 May 2021 (UTC)[reply]