Wikipedia:Bots/Requests for approval/RscprinterBot 2


 * The following discussion is an archived debate. Please do not modify it. To request review of this BRFA, please start a new section at WT:BRFA. The result of the discussion was Symbol delete vote.svg Denied

RscprinterBot 2
Operator:

Time filed: 12:32, Saturday January 14, 2012 (UTC)

Automatic or Manual: Automatic supervised

Programming language(s): AWB

Source code available: No

Function overview: Changes decade endings on pages to without an apostrophe, if it has one.

Links to relevant discussions (where appropriate):

Edit period(s): Daily

Estimated number of pages affected: 2 or 3 epm

Exclusion compliant (Y/N): probably not, but I don't know as I haven't run AWB in bot mode before

Already has a bot flag (Y/N): Y

Function details: Uses AWB to find decades (1920's, 1960's etc) and takes out the apostrophe if present, ie 1920's → 1920s. Just the same as.

Discussion
Just a precaution, are there any articles that deal with units, years or punctuation in case they discuss this convention or something? However, if you do supervise each edit (check it some time after it's done), then it should be O.K. Also AWB should be exclusion compliant even in bot mode. — HELL KNOWZ  ▎TALK 12:36, 14 January 2012 (UTC)
 * I am supervising each edit, and will skip unless it clearly needs it, like the years. Units and things will be left if they don't need it, and thanks for the clarification for the exclusion compliancy. Very quick reply!  Rcsprinter  (Gimme a message)  12:40, 14 January 2012 (UTC)
 * I'm wondering why you need a bot flag if you only do 2-3 epm and check each edit? Do you want the bot to do them in batches, then check the edits later or something? — HELL KNOWZ  ▎TALK 12:42, 14 January 2012 (UTC)
 * 2/3 epm is only a rough guestimate, could get a lot quicker, and that is what I want, big batches and check later, otherwise I would not need a flag. That's what's happening.  Rcsprinter  (gas)  12:44, 14 January 2012 (UTC)
 * Can you list out all changes this makes? Like the 1980's, the 1980-ies, the 1980s' . Are there any false positives there could be? —  HELL KNOWZ  ▎TALK 12:50, 14 January 2012 (UTC)
 * That's all, 0's to 0s. It will go through the 20s, 30s, 40s, 50s, 60s, 70s, 80s, 90s and then round again and again. False positives? I suppose it could do a unit or do something like "1950's great storm" to "1950s great storm" where the apostrophe is needed to denote that it happened in 1950 instead of the whole decade, but those can be easily rectified.  Rcsprinter  (talk to me)  12:53, 14 January 2012 (UTC)
 * This does look like a case of WP:CONTEXTBOT. This should need some proper empirical investigation into false positives. If cases need to be corrected, then it's best to not cause them in the first place. — HELL KNOWZ  ▎TALK 12:56, 14 January 2012 (UTC)

Like I've said, the false positives probably won't happen so they won't need to be rectified, but you asked so I'm just telling you the worst.  Rcsprinter  (talk to me)  12:58, 14 January 2012 (UTC)
 * How do you find the list of pages that need this corrected? — HELL KNOWZ  ▎TALK 13:03, 14 January 2012 (UTC)
 * In the 'make list' thing in AWB, I type the decade I'll be doing, with the apostrophe, and then set it going with the find and replace.  Rcsprinter  (Gimme a message)  13:06, 14 January 2012 (UTC)
 * A task like this would probably be best done on a database dump. And for approval purposes, it would mean you can actually make a full list of all pages and all changes without needing to scan each page at a time. From search results, I see that this occurs less than 1% of the time when you search for "1980's" for example. — HELL KNOWZ  ▎TALK 13:10, 14 January 2012 (UTC)
 * That is all a bit complicated for me; I'm fine with it the way it is.  Rcsprinter  (Gimme a message)  13:13, 14 January 2012 (UTC)
 * Well, personally, I'd want to see some assurance there are no false positives. That said, This isn't to say I'm pro approval or the method you use, I just want to see some edits for now. It'd be good if you say say how many pages AWB read and how many actually edited or had the issue. —  HELL KNOWZ  ▎TALK 13:15, 14 January 2012 (UTC)
 * Just to chip in that I agree with H3ll's thinking here. False positive rate is the thing to watch here. - Jarry1250 [Deliberation needed] 20:01, 14 January 2012 (UTC)
 * . Be aware that some of task one may be mixed in there.  Rcsprinter  (deliver)  20:44, 14 January 2012 (UTC)

4 false positives from 48 edits is not an acceptable false positive rate. Additionally, I'd like to point out that I specifically clarified if you would supervise each edit to which you said you would. It doesn't inspire confidence that you didn't even verify your trial edits. — HELL KNOWZ  ▎TALK 09:53, 15 January 2012 (UTC)
 * -- this is inside a title of a work
 * -- these are inside titles of works
 * -- this is not a year
 * -- this is inside a title of a work
 * What does it matter if it's in the title of a work? It's still only correcting the apostrophe. The Billboard Hot 100 one was 1 false positive. I was checking every edit. What is the next step from here?  Rcsprinter  (talk to me)  14:19, 15 January 2012 (UTC)
 * If you have checked the edits, why haven't you corrected at least the BB Hot 100 one? And why would it be okay to change the title of a work? You will need to show consensus for such a change by bot. — HELL KNOWZ  ▎TALK 15:43, 15 January 2012 (UTC)
 * I have checked the edits, and corrected the BB Hot 100 one. I suppose if the title of the work was spelled that way we'd have to leave it sic, I don't have any sort of consensus for that. Were the rest of the edits OK? Have you got any sort of verdict?  Rcsprinter  (gossip)  16:39, 15 January 2012 (UTC)
 * You corrected it after I pointed it out. If you insist that you had seen it, may I ask why you left it? As per the BRFA itself, I'm not going to make any calls myself yet and would give this at least an extended trial. — HELL KNOWZ  ▎TALK 16:56, 15 January 2012 (UTC)
 * I was too busy to correct it at the time, and I've figured out a way to fix the bug: now it will only do decades but not units or titles etc. I think I'm ready for an extended trial.  Rcsprinter  (gossip)  20:45, 15 January 2012 (UTC)

I don't know how you will reliably avoid changing stuff in quotes and such, but since this is "supervised", then potential approval would require you verify edits. Anyway, extended trial on a larger set. — HELL KNOWZ  ▎TALK 12:43, 16 January 2012 (UTC)

Five edits in a row. Gimmetoo (talk) 21:23, 16 January 2012 (UTC)
 * - in title
 * - in title, and not a decade
 * - not a decade; undone by operator
 * - not a decade
 * in title; undone by operator; the title of the book appears to use 70s rather than 70's, so the bot edit appears to have been correct.


 * . I think there may be 202 edits there, but only because I wasn't keeping my eye on AWB when it finished. I know that there were few false positives in there, but they can all be corrected and as I wasn't expecting to get another 200 edits I have nearly finished the task anyway, but would still like approval as it will be an ever-ongoing thing.  Rcsprinter  (converse)  16:19, 17 January 2012 (UTC)

This level of false positives is unacceptable for a fully automated bot, especially considering it is interacting with the article space. If you want to run this task, do it manually under your main account and please check each edit -- Chris 04:14, 21 January 2012 (UTC)


 * The above discussion is preserved as an archive of the debate. Please do not modify it. To request review of this BRFA, please start a new section at WT:BRFA.