Cannabis Ruderalis

Genesis filter

- Otterathome (talk) 22:32, 22 March 2009 (UTC)

We already have one filter dedicated to Tile join. I think I am going to defer to others on this one. I worry though that those suggested phrases are dangerously specific so that the vandal could avoid them with only a tiny change in behavior. Dragons flight (talk) 01:33, 23 March 2009 (UTC)
Done as Special:AbuseFilter/83 - in log-only mode now for evaluation. This is exactly the kind of pattern abuse the abuse filter was meant to prevent. Mr.Z-man 02:02, 23 March 2009 (UTC)
Just to clarify, my point wasn't that we shouldn't go after him, it was that someone better acquainted with the issue might have a better idea how to do that effectively. Dragons flight (talk) 02:13, 23 March 2009 (UTC)
Earth, Age of the Earth and Big Bang should also be included as these are also popular targets for this type of vandalism. Email me the code when you are done so I can suggest possible improvements.--Otterathome (talk) 11:09, 23 March 2009 (UTC)
It just got bypassed recently by Keet Onn (talk · contribs · deleted contribs · nuke contribs · logs · filter log · block user · block log)
Though I am happy to say the Genesis vandal filter did get two of the earlier accounts. I would suggest making the filters more sensetive, and consider making a separate filter for the Evolution page, as this appears to be the primary target.--Otterathome (talk) 19:12, 27 March 2009 (UTC)

Filter name

  • Task: Warm users leaving an unused formatting chunks in articles. (If you want to be sophisticated, you can exclude it when they're in nowiki tags or the sandbox.)
  • Reason: '''Bold text''', == Headline text ==, <gallery> Image:Example.jpg|Caption1 Image:Example.jpg|Caption2 </gallery>: It's extremely easy for even experienced users to accidentally hit one of those blue buttons above the edit window and leave a piece of syntactic detritus lying around. This is almost always either a test or a mistake, so it should prompt only a warning. Or would that be too much lag for the trouble?

- HereToHelp (talk to me) 03:22, 30 March 2009 (UTC)

Such a filter was one of the first ones added, Special:AbuseFilter/18. Dragons flight (talk) 03:32, 30 March 2009 (UTC)
Okay, thank you. --HereToHelp (talk to me) 17:11, 30 March 2009 (UTC)

chishtyajmersharif.com

  • Task: Disallow additions of chishtyajmersharif.com moinuddinchishty.com duagobollywood.com nawazish-e-khwaja.com chishtisufi.com meregharibnawaz.com byaagharibnawaz.com to articles.

- MER-C 08:47, 26 March 2009 (UTC)

Done. Special:AbuseFilter/107. I also added 4shared.com from the request lower on this page. Dragons flight (talk) 09:01, 26 March 2009 (UTC)

Reference compliance filter

(This is a revised re-sumbission of Wikipedia:Abuse filter/Requested#Notability filter after this discussion over at WP:BOTREQ.)

  • Task: Check if newly created articles by new users (under a certain threshold of edit count or account age) has multiple references external references (accounting both for inline citations and simply external links), and if not, bring up a warning beforehand about our verifiability policy before the posting is made.
  • Reason: Many new users are not aware of our core policies in regards to notability and verifiability, specifically the requirement for multiple, non-trivial sources, we tag sometimes hundreds of articles along the lines of "Jon is cool..... the end" for speedy deletion each week. This could possibly catch obvious articles of this type and educate the new contributor.

-Senseless!... says you, says me 16:43, 23 March 2009 (UTC)

We have CSD for lack of notability within certain classes, but I seem to recall that lack of references has been explcitly rejected as a CSD criteria. The task as proposed feels awfully bitey to me, and likely to hit a lot of people that mean well. I'd be more open to targeting the ultrashort articles, say new articles (not redirects) having less than 100 characters or something. Asking a new user to make their article longer should be more approachable than asking them to figure out our referencing system. Dragons flight (talk) 17:54, 23 March 2009 (UTC)
WP:BITE was brought up over at WP:BOTREQ. I never thought of asking new users to reference their contributions to be bitey, but that seems to be what consensus is stating. That being said, I agree that targeting ultra-short articles and asking (i.e. a warning message before saving changes) that they be legenthed might be a more realistic approach. -Senseless!... says you, says me 19:11, 23 March 2009 (UTC)
I don't spend much time around new articles, and basically pulled 100 characters out of the air. Before I turn this on, I am wondering if anyone has a better insight into what the threshold for "too short to be useful" should be? Dragons flight (talk) 23:14, 23 March 2009 (UTC)
I personally would have gone with 200, but I'm not sure what consensus would be on that. -Senseless!... says you, says me 04:52, 24 March 2009 (UTC)
I've gone with 200 for now after browsing the new pages log and finding that looked reasonable. Special:AbuseFilter/98. Dragons flight (talk) 17:28, 25 March 2009 (UTC)
  • It depends entirely on how you ask if requesting references is bitey. It is certainly less bitey than nominating someone's work for deletion. The downside is that people will make up bad references to get it through the filter... - Mgm|(talk) 23:30, 23 March 2009 (UTC)

Bad Chart filter

  • Task: Any edit to articles in Category:Albums or Category:Singles (including all subcategories) that includes any of the following phrases:
  • United World Chart
  • Media Traffic
  • Bulgaria Single
  • Bulgarian Single
  • Brazil Hot 100
  • Brasil Hot 100
  • Hot 100 Brazil
  • Hot 100 Brasil
  • Brazil Single
  • Brasil Single
  • Brazil Album
  • Brasil Album
  • Lithuanian Single
  • Lithuanian Airplay
  • MYX Hit Chart
  • MYX Daily Top 10
  • Pinoy MYX Countdown
  • Global Top 40
  • Poland Top 50
  • Poland National Top 50
  • Chile Top 40
  • Argentina Top 40
  • Euro 200
  • World Dance
  • World Trance
  • World Latin
  • World Jazz
  • World Soundtracks
  • OST Top 20

The edit should be blocked, with a warning message that points the editor at WP:BADCHARTS.

  • Reason:This would block addition of the most commonly added bogus charts to album and single articles. Look at my contribution history, and you can see what a massive task it is to undertake manually. If people are jittery about blocking with that comprehensive of a filter, I'd settle for a warning and a log to evaluate the results with an eye towards converting it into a blocking filter after a successful trial.—Kww(talk) 23:29, 22 March 2009 (UTC)
I'm willing to try this with a close eye on load, but the complexity of it may be too much. Dragons flight (talk) 23:54, 22 March 2009 (UTC)
If it is too computationally intensive, the Brazil, Bulgaria, and United World Chart filters combined will get about 75% of the effect.—Kww(talk) 00:00, 23 March 2009 (UTC)
Special:AbuseFilter/81. We reserve the right to modify or disable this in the future due to load, but for the moment it will log non-sysop edits to articles containing the words "single", "album", or "song" (not going to mess with categories) where any of the chart titles you identified are included in the edit (case insensitive). Dragons flight (talk) 00:54, 23 March 2009 (UTC)
Setting the filter to log will be fine, but the disallow flag should never be set. The abuse filter isn't supposed to make content decisions. --Conti| 00:03, 23 March 2009 (UTC)
I don't really see the difference between XlinkBot reverting one of these edits (if the editor has had the courtesy to reference his edit) and the Abuse filter blocking the edit when the editor was so sloppy that he not only added an unreliable source, but he didn't even add a reference for it.—Kww(talk) 00:40, 23 March 2009 (UTC)
The bot reverts and leaves a message at the reverted user's talk page (at least I hope so), the filter does not. You can't see from the edit history whether an edit to an article was blocked, and once a flag is set to disallow, I doubt a lot of people will check the abuse log for false positives. The abuse filter was created to deal with Grawp and the like, not to enforce content decisions. Others might disagree, but I think we should take things very slowly here. --Conti| 00:58, 23 March 2009 (UTC)
Yes, I can't see this ever going higher than Warn, and it will probably sit at log for quite a while to check for false positives in that massive regex before even doing that. Dragons flight (talk) 01:06, 23 March 2009 (UTC)
I'll try to persuade you to make it take stronger action after you see how frequently it will fire, and what a pervasive problem this is. For now, it will hopefully let me get my watchlist down to the sub-4000 article count again, and to not spend my time filtering through search results looking for these things. The first hit was a valid hit, and I've already removed the bad chart from the article.—Kww(talk) 02:20, 23 March 2009 (UTC)
One tweak I can suggest based on experience with searches: if looking at the categories is impractical, then you might want to ignore articles that contain the word "tennis". Bulgarians play tennis, and do quite well in singles matches.—Kww(talk) 02:31, 23 March 2009 (UTC)
No Bulgarian tennis players, check. Dragons flight (talk) 02:39, 23 March 2009 (UTC)

Some false triggers when I checked this morning:

  • [7] Tripped on a user adding an instructional comment not to add any of the charts listed in the filter. Not sure if it's feasible to ignore text in comments.
  • [8] Tripped on an article about a soccer player.
Kww(talk) 12:20, 23 March 2009 (UTC)

A negative that I don't understand: inserted Bulgarian chart, but didn't trip filter.—Kww(talk) 03:04, 28 March 2009 (UTC)

This one, too. I bet it's the reference brackets screwing up the match: he's hiding the redlink by making it the [[Bulgaria]][[Singles]][[Chart]].—Kww(talk) 03:29, 28 March 2009 (UTC)
Should be fixed. Dragons flight (talk) 05:16, 28 March 2009 (UTC)

I've reviewed several dozen of the logged matches and concluded this generates too many false positives in its current form to justify using an automated warning. So you can:

  1. Keep it active in its current log-only state and use that log to manage the problem by hand.
  2. Suggest changes to reduce the false positive rate to a reasonable level for warning.
  3. Keep this as is, and suggest the addition of a narrower rule that is unlikely to generate false positives with the intent of creating something that could be used for a warning.

Given that we only see ~20 hits per day, I'd suggest that 1) is probably the best solution. Keep it as is and have a group of you use it to target the problems (and the problem users) as needed. Dragons flight (talk) 01:47, 30 March 2009 (UTC)

I can only think of two ways to drop its FP rate: use the article category (which is apparently infeasible) or only apply the filter when the word "Chart" is in the section title where the text appears. If neither of those is feasible, just leave it like it is. It has cut my workload considerably, and doesn't make me rely on massive watchlists and post-mortem searches.—Kww(talk) 02:25, 30 March 2009 (UTC)

Wallflowers98 Filter

  • Task: Prevent
<nowiki>"The Trick" can be heard here:</nowiki>

<small><nowiki>http://www.4shared.com/file/91568176/551b87d3/The_Trick_320_kbps.html</small></nowiki>

from being added into the article namespace.

  • Reason: To prevent User:Wallflowers98 and his socks for using wikipedia as a gateway to illegally hosted files.

- — dαlus Contribs 23:23, 22 March 2009 (UTC)

So the person is using nowiki to continue referencing 4shared despite it being blacklisted? Dragons flight (talk) 23:44, 22 March 2009 (UTC)
It seems that way.— dαlus Contribs 23:45, 22 March 2009 (UTC)
Since it apparently bypasses the blacklist, is there ever a reason for new users to be adding nowiki'd urls to articles? Dragons flight (talk) 23:49, 22 March 2009 (UTC)
I wouldn't believe so, are you suggesting that instead we just create a filter stopping people from no-wiking urls into article text?— dαlus Contribs 23:53, 22 March 2009 (UTC)
Yes, unless there is some legitimate use for that behavior. Dragons flight (talk) 23:55, 22 March 2009 (UTC)
Well, I can't think of any, so... — dαlus Contribs 23:58, 22 March 2009 (UTC)
Special:AbuseFilter/82, testing at log-only. Catches new users adding urls in nowiki tags. Dragons flight (talk) 01:07, 23 March 2009 (UTC)
Does moving it to this section mean that it will be changed to disallow rather than log, or?— dαlus Contribs 08:23, 23 March 2009 (UTC)
It means a functional filter exists and the upper part of the page is too cluttered.  :-) Decisions about log -> warn -> disallow, etc. will be based on whether we get false positives, etc. Dragons flight (talk) 00:04, 24 March 2009 (UTC)
  • No, there are legitimate uses of putting links in nowiki tags so we won't be disallowing it anytime soon. Anyone can visit the Special:AbuseLog and revert edits caught by filter 82. With enough eyes the good and bad edits can be separated. Just logging alone makes it harder to fly under the radar with this. - Mgm|(talk) 23:33, 23 March 2009 (UTC)
    • Can you give an example of why a nowiki'd link should be in an article. I honestly can't think of one. Dragons flight (talk) 00:01, 24 March 2009 (UTC)
      • So far, the only things caught by filter 82 are vandals trying to bypass the link blacklist.— dαlus Contribs 00:33, 24 March 2009 (UTC)
        • This is the only false positive, can the filter be adjusted?— dαlus Contribs 00:54, 24 March 2009 (UTC)
          • Actually, I already fixed that by tightening the regex hours ago. Dragons flight (talk) 00:55, 24 March 2009 (UTC)
      • I have to concur that I can't come up with a good argument for a nowikied URL in article space. I can come up with them for talk space, but not for article space.—Kww(talk) 01:12, 24 March 2009 (UTC)
        • Agreed. The only plausible reason to nowiki something in the article space is to try to avoid our blacklist. --slakrtalk / 05:27, 24 March 2009 (UTC)
          • It looks like we have a few more false positives, can someone adjust it with regards to ref tags?— dαlus Contribs 04:22, 26 March 2009 (UTC)
            • Can we enable this filter to disallow edits yet?— dαlus Contribs 06:25, 22 April 2009 (UTC)
            • I would concur with setting this to disallow as long as it doesn't trip on the MediaWiki talk and Wikipedia talk namespaces. Stifle (talk) 17:46, 26 April 2009 (UTC)

Goatse...

Task: Can we try to make a filter to disallow popular ascii art representations of Goatse...noting some vandalism that occurred on the main abuse filter page from a slightly angry user? ViperSnake151 17:27, 19 March 2009 (UTC)

  • Such a filter was already written, disabled for performance reasons, but now re-enabled after some adjustments. -- Gogo Dodo (talk) 18:46, 19 March 2009 (UTC)
    • In Special:Abusefilter/65 I've also installed a warning for edits that contain large amounts of consecutive whitespace. This is one frequent indicator of ASCII art, but it also triggers on a variety of vandalism and when editors try to add pre-formatted text copied from outside Wikipedia with lots of spaces for formatting. Dragons flight (talk) 21:43, 19 March 2009 (UTC)

Montalbano-related spam/vandalism

There's a cross-wiki vandal who's spamming certain words to various articles. See these IPs' edit histories:

This is a recurring problem on nl.wikipedia; see the edits made by various IPs summarized on this page:

I suggest filtering for the phrases used by this spammer. I'd do this myself but it might be best if someone more familiar with the filter handled this. --A. B. (talk • contribs) 21:34, 7 April 2009 (UTC)

Someone beat you to it. Filter #44. —Wknight94 (talk) 22:05, 7 April 2009 (UTC)

Redirects to Barack Obama

- Zzyzx11 (Talk) 02:55, 29 March 2009 (UTC)

  • I am still learning how to use the tool, so I apologise for not doing this on my own first :-) Zzyzx11 (Talk) 03:09, 29 March 2009 (UTC)
    • Ok I created Special:AbuseFilter/116. If anybody would like to correct my syntax, feel free. Zzyzx11 (Talk) 03:17, 29 March 2009 (UTC)
      • And you get the bonus prize for prohibiting new users from making any redirects at all. Dragons flight (talk) 03:39, 29 March 2009 (UTC)
        • Test filters before setting them on Disallow. Thank you. Prodego talk 16:35, 29 March 2009 (UTC)

Mass symbol insertion

  • Task: Filter will prevent the insertion of the string "Ǣ ǣ ǖ ǘ ǚ ǜ Ă" into content namespaces by non-admins and into any namespace for a user with under 100 edits.
  • Reason: People who don't know better copy the whole symbols table from the edit tools below the window. Such vandalism blends into long articles and can stay there for a long time [9], there would be no valid reason to insert the entire character set at one time into a content page.

- MBisanz talk 00:14, 4 April 2009 (UTC)

I've modified Filter 18 accordingly. --Conti| 19:42, 4 April 2009 (UTC)

Leave a Reply