393:. When you say "rather than just adding them to general fixes and letting people do them when they make more significant edits?", that isn't really what happens with AWB in relation to CHECKWIKI... you use the lists there to generate a list for AWB and then basically just run the general fixes through it. However, doing this still takes quite a bit of time, so the backlogs there are still building up. Since many of these edits don't require humans to actually look at the article too much, it would be a huge timesaver to have a bot do them. For example, about a week or two ago I ran AWB on a list of about 300 or 400 article which had the CHECKWIKI error "Link equal to linktext". While doing so, all that I really did is glance over what general fixes were done to make sure that there weren't errors and clicking "save". This still took probably close to an hour and a half. If a bot had been doing this, I could have been working on some of the other problems which can't be done as easily. In essence, right now those lists are fixed by just using AWB's general fixes without any "more significant edits" being done at the same time, but a bot could handle it faster. These edits also appear on watchlists and in recent changes when they really shouldn't need to be.
330:
AWB or in the AutoEd script, so that it can be ensured that the change can be made reliably. To answer your second question, if anything is causing an at all unacceptable number of false positives I will not hesitate to stop the bot from making that change, regardless of my own feelings in the particular case. A false positive is worse than no edit at all, so deactivating a problematic change is the only logical thing to do. Of course, deactivating due to a single false positive doesn't really make sense in my mind, but if there are multiple complaints or concerns then that change should
809:
922:
434:
47:
206:. It would also do other cleanup at the same time... header name improvements ("Weblinks" → "External links", for example), link simplification and cleanup, and adding bullet points to external links. When running, anyone should be able to shut the bot off by posting on its talk page. I'd only run DrilBot when I'm around so that I can deal with any errors quickly. –
364:. What fixes, other than AWB general fixes will it be making? Also, bots doing solely AWB general fixes are denied. What makes the additional fixes significant enough that this needs to be done quickly with a bot, rather than just adding them to general fixes and letting people do them when they make more significant edits?
201:
to find and repair common errors, such as
Unicode control characters, bold text in and colons at the end of section headings, missplaced categories and interwikis, and links to the current article. This would be done using the basic "general fixes" of AWB, which repair some of the errors, and custom
270:
Don't worry
Drilnoth, I'm sure you know the scale of the problem! (It was the bot people around here I was primarily addressing, ;) ). Well, I've never found much room for false positives on some of the simpler tasks, as Drilnoth says. Just slightly too close to trial you myself, but it's shouldn't
401:
does some of the fixes, but I'm not sure if it should be since I can't find a BRFA for that task. I feel that having an approved bot to help manage the CHECKWIKI lists would make it much easier to maintain since then human editors could focus on the things which bots can't do rather than trying to
329:
I certainly don't think that there would be anywhere near 5% false positives. To answer your first question, yes. I have used AWB's general fixes a lot as can be seen in my contributions. Additionally, any custom-added regular expressions would first be tested out either with manual supervision in
405:(also, as a side note, it looks like Lightbot's controversy was because it changed date formatting; DrilBot shouldn't be doing anything that could cause that much controversy since all of the fixes would be to known errors, not things like date-formatting which can vary from article to article). –
318:
The
Knowledge (XXG) community tends to be extremely intolerant of "bot false-positives", meaning automated changes that need to be reverted. A 5% false positive rate would be far too much, especially with a heavy volume of edits. With that in mind, I have some questions. One, are all these fixes
247:
I am aware of the sheer number of errors there and how many more there are each day; I'd do my best to have DrilBot running more or less constantly while I'm editing to work on some of the backlogs which can be done automatically, so that human editors can focus on doing the things that can't be
738:
There were some cases where the error detected at CHECKWIKI couldn't be fixed; in some cases DrilBot then skipped the article entirely and in some cases it made other improvements. This isn't a problem per se, just saying that not all articles that it edits will still have the error detected at
627:, but it wont fix all of them. (I once did the other similar report). There are other reports, e.g. #7 where AWB can fix many, but you will still need to do the remaining ones manually. As for report #64, I think it should be de-activated, there isn't really a point in fixing mainly "]".
630:
Keep in mind that it's less risky to use a regex that fixes problem #xx through a list of articles with problem #xx than the same applied to all articles. You might want to look at the various general fixes and make sure you really want all of them applied automatically (see
396:
Since DrilBot would only use the lists created by CHECKWIKI, almost every article that it edits would have an error which it could fix, so there won't be a ton of edits that just fix things like whitespace or reference order, which I agree would be kind of useless. Right now
662:
The way that I see it, if a general fix doesn't have many false positives it is beneficial to apply it at the same time as the CHECKWIKI edits. If a particular change seems prone to false positives I'll deactivate it the way you mentioned, which I already did with date
319:
things you've previously used AWB to fix under your own account, manually assisted? And two, if enough people complain about a certain class of false-positive, are you willing to suspend that function, even if you feel it's doing far more good than harm? –
778:
and it should be fixed in the SVN builds (which I plan to try and download if DrilBot is approved, to have the latest version) and the next full release. Until I can update my version of AWB, I'll have the bot skip and page containing either of those
233:(n.b. I am a member of aforementioned WikiProject.) I suggest anyone interested looks at the main project page, to try to grapple with the scale of the problems on merely a day-to-day basis (new article feed etc.). Would you be prepared to actually
521:
rare occurence; there was just a set of almost identical articles and almost all of them had this error. I've never seen this particular code before, and once these article are manually fixed I doubt that it should ever really come up
290:
tags to the top of the page, but that is uncommon enough and really is an error with the article which just isn't fixed correctly that I think the benefits would far outweigh the small number of false positives that the bot would get.
580:
I can't speak for
Quadell, but I would advise you make the most of the trial - perfect the regexes, get some edits done. Record your fixes (and what you could fix) and report back here at the end. Good going so far, by the way. -
495:: Changed REFERENCES to ReferenceS, with the capital "S". Both the all caps and the some caps versions are wrong, so it was just a change from one wrong version to a different (maybe slightly less) wrong version.
513:: These four have very poorly done DEFAULTSORTS (specifically, {{DEFAULTSORT:{{BASEPAGENAME}}}}) and DrilBot got confused. These DEFAULTSORTS really shouldn't be there in the first place (they make
386:
1) Non-AWB fixes which I'd add would include things like better "unicodifying", removal of problematic
Unicode control characters, and some more template/link cleanup as they are tested in AutoEd.
402:
manage the whole list. If there aren't many false positives and the bot is flagged to prevent its appearance on watchlists/recent changes, I don't really see how this could be problematic.
963:
485:: Date changes where the dates were part of a book name. I fixed this error to prevent its occurrence in the future at about 16:30 UTC by disabling that particular AWB general fix.
605:, I think it's worth mentioning in the edit summary which list the bot is currently processing. This can help checking what was fixed, even if other general fixes are applied too.
280:
Ah, gotcha. Yes there certainly will be some false positives... I won't deny that. AWB's general fixes have some changes which get some edits incorrect, e.g. moving all
690:
Some of the unbalanced brackets issues (both square and curly) are fixed automatically (approx 20-30% in my experience), but most do indeed need human supervision. -
334:
be deactivated. Bots may be run and maintained by just one user, but I feel that their actions should really be determined by the community, not just one person. –
742:
In addition to the false positives above (which were weird situations or which have now been fixed), there were three other pages where DrilBot caused an error:
632:
510:
506:
502:
646:
I can include which list is being processed, although sometimes I might just omit that part of the summary if I'm doing a short list (like 15-30 items).
21:
639:
I think you did a lot of good work and WP:CHECKWIKI and I'd happily back any proposal that helps you. -- User:Docu 22:42, 8 May 2009 (UTC)
731:
I'm going to be away for a few hours, so here's my report on the 500+ edits that DrilBot made yesterday (I checked all of them for errors):
498:
271:
take too long to get the trial to make sure this is fertile ground for a bot, considering the possibility for false positives. -
747:
608:
If you come across general fixes that need improvement, e.g. defaultsort in the sample above, I think they should mentioned at
87:
659:
I had been under the impression that the "AWB assisted" problems required human attention to fully fix... is that not correct?
906:
866:
824:
792:
708:
677:
567:
535:
459:
414:
343:
300:
257:
215:
137:
82:
438:
Approved for trial (1 days). Please provide a link to the relevant contributions and/or diffs when the trial is complete.
202:
regular expressions after they have been tested to ensure a minimum of false positives using the assisted editing script
389:
2) The additional fixes themselves aren't what I feel makes this sort of bot needed, rather it's the massive backlog at
478:
The trial went almost perfectly; I'll just report the handful of false positives that I saw (and reverted or fixed):
117:
102:
751:
558:
Okay; I wasn't sure if you wanted me to give occasional reports or just lump it all together at the end. –
624:
602:
390:
383:(added some more indentation). There are a few questions here, and I'll do my best to answer all of them.
198:
166:
932:
914:
887:
874:
832:
800:
716:
694:
685:
585:
575:
553:
543:
467:
444:
422:
378:
351:
323:
308:
275:
265:
241:
223:
609:
910:
870:
828:
796:
712:
681:
571:
539:
463:
418:
347:
304:
261:
219:
154:
141:
883:
This seems fine to me. If there are no objections in the next few days, I'm inclined to approve. –
492:
97:
653:
617:
616:
If there are fixes that could be applied to all operations, I think it's worth suggesting them at
203:
194:
92:
902:
862:
820:
788:
774:
templates to the top of the article when they properly belonged in a section. I reported this at
704:
673:
649:
Of course; I just didn't mention this one because, as I said, it's an extraordinarily rare error.
563:
531:
455:
410:
369:
339:
296:
253:
211:
133:
775:
735:
Things went almost perfectly; there were just a handful of false positives as described below.
488:
77:
17:
851:
768:
743:
482:
894:
517:
on categorization), and I don't think that AWB was set up to handle it. This should be a
691:
582:
272:
238:
957:
898:
858:
816:
784:
700:
669:
559:
527:
451:
406:
366:
361:
335:
292:
249:
207:
129:
929:
884:
841:
758:
550:
441:
320:
284:
58:
656:, although most of what I'd suggest are already implemented in the next version.
46:
398:
440:
Let's get a feel for the sorts of changes we're talking about here. –
613:{{SUBST:BASEPAGENAME}} should probably be added to the general fixes)
635:
for the details). Smackbot probably doesn't use all of them either.
620:. This gives the developers feedback and helps develop the program.
942:
The above discussion is preserved as an archive of the debate.
193:
Using AWB and my own RegExp (after it has been tested using the
165:
Cleans up various common errors in articles using the lists at
625:
WP:CHECKWIKI#Square_brackets_not_correct_begin_(AWB_assisted)
633:
WP:AutoWikiBrowser/Custom_Modules#Customised_"General_Fixes"
783:
I didn't notice any other potentially problematic edits. –
652:
Most definitely; I've already mentioned a few things at
360:
Approving bots with super-generic "fixes"-type tasks is
112:
107:
72:
857:
problem should now be fixed thanks to an AWB update. –
948:
Subsequent comments should be made in a new section.
42:
Subsequent comments should be made in a new section.
964:Approved Knowledge (XXG) bot requests for approval
248:fixed by bot, like incorrect brackets and ISBNs. –
175:Basically whenever I'm editing Knowledge (XXG).
36:The following discussion is an archived debate.
612:. This avoids repeating them. ( <nowiki: -->
754:. These were all instances where it moved the
549:Yes, you have 22 hours or so left to go. :) –
897:per Jarry1250's suggestion on my talk page. –
601:If you are working on the various lists from
526:Shall I continue the trial or stop for now? –
8:
511:Coat of arms of Victoria, British Columbia
507:Coat of arms of Victoria, British Columbia
503:Coat of arms of Victoria, British Columbia
197:), DrilBot would use the daily lists at
893:Thanks. I've posted about DrilBot at
7:
643:Thanks; I have a few comments here.
28:
499:Coat of arms of Whitehorse, Yukon
199:WikiProject Check Knowledge (XXG)
167:WikiProject Check Knowledge (XXG)
44:The result of the discussion was
920:
807:
432:
45:
806:Oops, hadn't seen this before:
748:Sport Club Corinthians Paulista
148:Automatic or Manually Assisted:
1:
980:
699:Ah; thanks for the info. –
362:somewhat frowned upon now
237:it, or just be around? -
945:Please do not modify it.
933:01:22, 12 May 2009 (UTC)
915:16:05, 10 May 2009 (UTC)
888:15:34, 10 May 2009 (UTC)
875:15:05, 10 May 2009 (UTC)
39:Please do not modify it.
833:18:15, 9 May 2009 (UTC)
801:13:32, 9 May 2009 (UTC)
717:13:33, 9 May 2009 (UTC)
695:10:08, 9 May 2009 (UTC)
686:01:24, 9 May 2009 (UTC)
586:17:57, 8 May 2009 (UTC)
576:17:54, 8 May 2009 (UTC)
554:17:04, 8 May 2009 (UTC)
544:16:48, 8 May 2009 (UTC)
468:15:53, 8 May 2009 (UTC)
445:15:17, 8 May 2009 (UTC)
423:14:13, 8 May 2009 (UTC)
379:06:01, 8 May 2009 (UTC)
352:19:52, 7 May 2009 (UTC)
324:19:06, 7 May 2009 (UTC)
309:16:51, 7 May 2009 (UTC)
276:16:44, 7 May 2009 (UTC)
266:16:36, 7 May 2009 (UTC)
242:16:34, 7 May 2009 (UTC)
224:16:27, 7 May 2009 (UTC)
752:Undrafted sportsperson
623:AWB might fix some of
179:Already has a bot flag
22:Requests for approval
155:Programming Language
18:Knowledge (XXG):Bots
727:Report after trial
163:Function Overview:
450:Can do; thanks. –
191:Function Details:
971:
947:
924:
923:
856:
850:
846:
840:
811:
810:
773:
767:
763:
757:
744:Real Madrid C.F.
483:Kenneth W. Royce
436:
435:
377:
289:
283:
159:AutoWikiBrowser
49:
41:
979:
978:
974:
973:
972:
970:
969:
968:
954:
953:
952:
943:
921:
854:
848:
844:
838:
813:Trial complete.
808:
771:
765:
761:
755:
433:
365:
287:
281:
231:
173:Edit period(s):
123:
62:
37:
26:
25:
24:
12:
11:
5:
977:
975:
967:
966:
956:
955:
951:
950:
937:
928:Looks good. –
918:
917:
882:
880:
879:
878:
877:
781:
780:
740:
736:
729:
728:
724:
723:
722:
721:
720:
719:
666:
665:
664:
660:
657:
650:
647:
637:
636:
628:
621:
614:
606:
598:
597:
593:
592:
591:
590:
589:
588:
524:
523:
496:
489:Balázs Megyeri
486:
476:
475:
471:
470:
430:
429:
428:
427:
426:
425:
403:
394:
387:
384:
355:
354:
316:
315:
314:
313:
312:
311:
230:
227:
122:
121:
115:
110:
105:
100:
95:
90:
85:
80:
75:
73:Approved BRFAs
70:
63:
61:
56:
55:
54:
32:
30:
27:
15:
14:
13:
10:
9:
6:
4:
3:
2:
976:
965:
962:
961:
959:
949:
946:
940:
939:
938:
935:
934:
931:
927:
916:
912:
908:
904:
900:
896:
892:
891:
890:
889:
886:
876:
872:
868:
864:
860:
853:
843:
836:
835:
834:
830:
826:
822:
818:
814:
805:
804:
803:
802:
798:
794:
790:
786:
777:
770:
760:
753:
749:
745:
741:
737:
734:
733:
732:
726:
725:
718:
714:
710:
706:
702:
698:
697:
696:
693:
689:
688:
687:
683:
679:
675:
671:
667:
663:reformatting.
661:
658:
655:
651:
648:
645:
644:
642:
641:
640:
634:
629:
626:
622:
619:
615:
611:
607:
604:
600:
599:
595:
594:
587:
584:
579:
578:
577:
573:
569:
565:
561:
557:
556:
555:
552:
548:
547:
546:
545:
541:
537:
533:
529:
520:
516:
515:no difference
512:
508:
504:
500:
497:
494:
493:Balázs Vattai
490:
487:
484:
481:
480:
479:
473:
472:
469:
465:
461:
457:
453:
449:
448:
447:
446:
443:
439:
424:
420:
416:
412:
408:
404:
400:
395:
392:
388:
385:
382:
381:
380:
376:
375:
373:
368:
363:
359:
358:
357:
356:
353:
349:
345:
341:
337:
333:
328:
327:
326:
325:
322:
310:
306:
302:
298:
294:
286:
279:
278:
277:
274:
269:
268:
267:
263:
259:
255:
251:
246:
245:
244:
243:
240:
236:
228:
226:
225:
221:
217:
213:
209:
205:
200:
196:
195:AutoEd script
192:
188:
186:
183:
180:
176:
174:
170:
168:
164:
160:
158:
156:
151:
149:
145:
143:
139:
135:
131:
127:
119:
116:
114:
111:
109:
106:
104:
101:
99:
96:
94:
91:
89:
86:
84:
81:
79:
76:
74:
71:
69:
65:
64:
60:
57:
52:
48:
43:
40:
34:
33:
31:
23:
19:
944:
941:
936:
925:
919:
881:
812:
782:
730:
638:
603:WP:CHECKWIKI
525:
518:
514:
477:
437:
431:
391:WP:CHECKWIKI
371:
370:
331:
317:
234:
232:
190:
189:
184:
181:
178:
177:
172:
171:
162:
161:
153:
152:
147:
146:
125:
124:
67:
50:
38:
35:
29:
610:WP:AWB/BUGS
779:templates.
739:CHECKWIKI.
332:definitely
229:Discussion
150:Automatic
113:rights log
103:page moves
926:Approved.
692:Jarry1250
668:Thanks! –
654:WP:AWB/FR
618:WP:AWB/FR
583:Jarry1250
273:Jarry1250
239:Jarry1250
126:Operator:
108:block log
958:Category
899:Drilnoth
859:Drilnoth
817:Drilnoth
785:Drilnoth
776:WP:AWB/B
701:Drilnoth
670:Drilnoth
560:Drilnoth
528:Drilnoth
452:Drilnoth
407:Drilnoth
336:Drilnoth
293:Drilnoth
250:Drilnoth
208:Drilnoth
130:Drilnoth
83:contribs
51:Approved
20: |
930:Quadell
885:Quadell
852:dablink
769:dablink
551:Quadell
442:Quadell
399:User:D6
321:Quadell
59:DrilBot
895:WP:VPM
750:, and
522:again.
509:, and
474:Report
204:AutoEd
235:watch
182:(Y/N)
88:count
16:<
847:and
837:The
596:Note
519:very
491:and
157:(s):
118:flag
98:logs
78:talk
68:BRFA
842:for
815:. –
764:or
759:for
374:man
367:Mr.
285:for
93:SUL
960::
913:)
909:•
905:•
873:)
869:•
865:•
855:}}
849:{{
845:}}
839:{{
831:)
827:•
823:•
799:)
795:•
791:•
772:}}
766:{{
762:}}
756:{{
746:,
715:)
711:•
707:•
684:)
680:•
676:•
574:)
570:•
566:•
542:)
538:•
534:•
505:,
501:,
466:)
462:•
458:•
421:)
417:•
413:•
372:Z-
350:)
346:•
342:•
307:)
303:•
299:•
288:}}
282:{{
264:)
260:•
256:•
222:)
218:•
214:•
187:N
169:.
144:)
140:•
136:•
911:L
907:C
903:T
901:(
871:L
867:C
863:T
861:(
829:L
825:C
821:T
819:(
797:L
793:C
789:T
787:(
713:L
709:C
705:T
703:(
682:L
678:C
674:T
672:(
572:L
568:C
564:T
562:(
540:L
536:C
532:T
530:(
464:L
460:C
456:T
454:(
419:L
415:C
411:T
409:(
348:L
344:C
340:T
338:(
305:L
301:C
297:T
295:(
291:–
262:L
258:C
254:T
252:(
220:L
216:C
212:T
210:(
185::
142:L
138:C
134:T
132:(
128:–
120:)
66:(
53:.
Text is available under the Creative Commons Attribution-ShareAlike License. Additional terms may apply.
