Wikipedia:Bots/Requests for approval/Ohms Law Bot 3


 * 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 neutral vote.svg Request Expired.

Ohms Law Bot 3
Operator:

Time filed: 02:23, Friday May 6, 2011 (UTC)

Automatic or Manual: Automatic unsupervised

Programming language(s): C#

Source code available: No

Function overview: Adjusts links within article content in accordance with WP:NOTBROKEN

Links to relevant discussions (where appropriate): WP:NOTBROKEN

Edit period(s): Continuous

Estimated number of pages affected: All(?)

Exclusion compliant (Y/N): Sure

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

Function details: for this task Ohms Law Bot will examine links within article content. If a link is piped, and the display portion of the link is the title of an article that is a redirect, while the target portion of a link is the article containing the content, then Ohms Law Bot will change the link to target the redirect page. This functionality would be in compliance with Redirect.

Discussion
Do you have a list of pages you wish to operate your bot on or are you just cycling allpages? Noom talk stalk 15:20, 9 May 2011 (UTC)
 * Actually, I haven't decided yet. I was considering just using random pages. — V = IR (Talk&thinsp;&bull;&thinsp;Contribs) 15:23, 9 May 2011 (UTC)
 * Per WP:BOTPOL, if you're requesting a lot of page content, you would be better off using a database dump. Noom  talk stalk 15:24, 9 May 2011 (UTC)
 * True, but I won't be requesting a lot of page content. It'll work on a single page at a time. — V = IR (Talk&thinsp;&bull;&thinsp;Contribs) 15:39, 9 May 2011 (UTC)
 * Are there many of these links that need to be fixed? Or you'll end up skipping through pages fairly quickly. Noom  talk stalk 15:45, 9 May 2011 (UTC)
 * In my experience manually editing pages there seems to be a lot of them. However, I don't actually have any statistics... It'll likely skip through a number of pages fairly quickly. I don't think that there's any way to avoid that, though. Note that I'm using the API though, so the server load shouldn't be very significant. — V = IR (Talk&thinsp;&bull;&thinsp;Contribs) 15:51, 9 May 2011 (UTC)
 * By the way, I anticipate adding other tasks eventually, so the whole "skipping though articles" issue shouldn't be much of a problem. This task likely won't run on every page, but it will be used. — V = IR (Talk&thinsp;&bull;&thinsp;Contribs) 17:05, 10 May 2011 (UTC)

OK I've reviewed this and I can't make any sense of what it is that the bot would actually do? Could you give a before/after example? Headbomb {talk / contribs / physics / books} 06:53, 10 May 2011 (UTC)
 * Sure, It's simple really. Say, for example, that someone places a link to Public-key cryptography on a page. Since exists as a redirect to Public-key cryptography, the link can (and should) point to the redirect page itself, not the target. So, the bot would change the link to Asymmetric Algorithms, which is what WP:NOTBROKEN advises us to do (in the majority of cases). — V = IR  (Talk&thinsp;&bull;&thinsp;Contribs) 17:03, 10 May 2011 (UTC)
 * I don't get it either. In your example you would change the displayed value of the link. How will the bot know if that fits the context/phrasing of the sentence, or if only the existing target value does? Rjwilmsi  15:13, 14 May 2011 (UTC)


 * I have to echo the above in not quite understanding this. For instance, "Smith worked on public-key systems and pioneered Public-key cryptography ." is the intended sentence. Changing this to "Smith worked on public-key systems and pioneered Asymmetric Algorithms. " implies a different thing. If anything, you would want to change it to "Smith worked on public-key systems and pioneered Public-key cryptography. "; but that is WP:NOTBROKEN. —  HELL KNOWZ  ▎TALK 16:09, 14 May 2011 (UTC)


 * I saw this coming, dammit. Unfortunately, I don't have a good answer for you guys other then "it won't be a problem". The reason is that... well, this isn't a single purpose bot program. There's a fairly significant back end to it, which performs some Natural language processing... AI is sorta what I do. Anyway, seeing as how I'm unwilling to share my code, and I anticipate that sharing some of the source would be an inevitable result of this line of questions, the only solution that I can think of is allowing a trial run (or outright denying the task... which I'll be fine with, but I hope doesn't happen.) — V = IR (Talk&thinsp;&bull;&thinsp;Contribs) 20:16, 15 May 2011 (UTC)
 * Here's something of an answer: The bot doesn't change the display text where there isn't a page matching the display text. For example, say that there's a sentence such as Smith worked on public-key systems and pioneered Public-key cryptography., and the pages "Public-key cryptography" exists, but the page for "Asymmetric Algorithms" doesn't (I realize that it does, but bear with me. This is an example, after all). If that's the case, the bot may change the target, but it will not change the display text. I've found out that, using this simple logic actually prevents the vast majority of the problems that you guys are bringing up here. I'd like to be clear: there will be occasionally "false positives" from this task. However, a handful of false positives (possibly) for every 100 changes that this task would create is an acceptable rate, to me. — V = IR (Talk&thinsp;&bull;&thinsp;Contribs) 20:38, 15 May 2011 (UTC)


 * As much as I'd like Wikipedia to be run by bots writing up perfect FAs, that's not going to happen. I don't think you will gather consensus for changing links like that and we'll have to deny the BRFA per WP:CONTEXTBOT. Bots have little to no sense of context. I will be very surprised if your code manages to outsmart that. But, per your own estimate, a handful false positives per 100 edits is more than CluebotNG is allowed, which is extremely useful, and even then had users arguing about false positives. — HELL KNOWZ  ▎TALK 20:50, 15 May 2011 (UTC)


 * LOL @ "Wikipedia to be run by bots writing up perfect FAs". :D — V = IR (Talk&thinsp;&bull;&thinsp;Contribs) 21:05, 15 May 2011 (UTC)

This is what Redirect deprecates. Indeed, it has less point; it does not change the visible text (Assymmetric algorithms in the example); it does not change where the reader winds up (Public-key cryptography, at the top of the page). It might be useful to take that link and make it a section redirect, but that's more than a bot can do.

So why? What benefit does this have besides the trivial one of making the redirect faster, which is not enough to jusitify the time required to process another edit of the linking page? Septentrionalis PMAnderson 18:06, 15 May 2011 (UTC)


 * The example does change the visible text: Public-key cryptography to Asymmetric Algorithms ; i.e. Public-key cryptography to Asymmetric Algorithms. That's the thing. A bot doing WP:NOTBROKEN would have been denied already. — HELL KNOWZ  ▎TALK 18:20, 15 May 2011 (UTC)
 * Thanks, sorry for the confusion. Oppose until Rjwilmsi's question is answered; this is guessing that the link will be better than the actual words written. Septentrionalis PMAnderson 19:29, 15 May 2011 (UTC)


 * Here's the sort of change that the BOT would make:
 * I made that change manually of course, but the logic is there for the bot to do exactly what I did myself. Actually, I could have the bot make the links go to page sections as well, using logic something along the lines of "if there's a section header on the target page matching the original target, point to that", but adding that to this task might be slightly controversial.
 * Also, please note that I'm philosophically opposed to bots "protecting" their edits (at least, for bots such as mine). This bot will not, and indeed can not, revert edits by others (at least, not intentionally). I know that this comment isn't really related to anything that has been brought up, but I hope that it'll... inspire confidence, somewhat? Allay some fears that might exist, I guess. — V = IR (Talk&thinsp;&bull;&thinsp;Contribs) 19:56, 15 May 2011 (UTC)
 * If this is restricted to See also sections, it's not rewriting prose, which is something. Is it?


 * That it cannot revert is positively useful; if you could send that code to some other editors on this page, they would become more useful contributors. Septentrionalis PMAnderson 02:06, 16 May 2011 (UTC)

A dry run trial for a concept proof. Make some link change reports with the surrounding prose included, so we can see how the bot would handle those cases and if the rewording is correct. — HELL KNOWZ  ▎TALK 06:23, 16 May 2011 (UTC)
 * OK, working on getting this done now. I need to change a couple of things first, and I'm suddenly really busy off-wiki, but... it'll happen. Thanks! I have to admit that I didn't foresee getting any approval for this, so you've caught me a tad bit flat footed here. Sorry. :) — V = IR (Talk&thinsp;&bull;&thinsp;Contribs) 18:15, 17 May 2011 (UTC)
 * Any progress?  MBisanz  talk 23:22, 12 June 2011 (UTC)
 * Some. I had to take some time off from development for a while, but I'm back at it now. I screwed up my local database though, so I'm in the process of recreating it, which is going to take a little bit of time. — V = IR (Talk&thinsp;&bull;&thinsp;Contribs) 23:33, 12 June 2011 (UTC)
 * Ok, sounds good. No hurry, I just like to make sure I'm not neglecting older requests.  MBisanz  talk 23:15, 19 June 2011 (UTC)
 * How is this request coming along? SQL Query me!  07:58, 3 August 2011 (UTC)

No response from the operator in 8 days, this subpage has not been edited by the operator in months. without prejudice. Please re-open this request whenever you would like. SQL Query me! 05:13, 11 August 2011 (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.