Wikidata talk:Abuse filter/Archive/2015

Most important pages

•  Done I suggest an abuse filter for the english labels and descriptions of the 10,000 most important items on Wikidata. These are well established english labels and descriptions, there is a 99.99% chance that any edit there is a vandalism. For example: Antoine Lavoisier (Q39607) and Franklin Delano Roosevelt (FDR). --Chris.urs-o (talk) 08:37, 3 January 2015 (UTC)

Discussion

@Chris.urs-o: How do you know it is top 10,000 item? Matěj Suchánek (talk) 08:46, 3 January 2015 (UTC)

Well, the beginning of Wikidata used the top 1,000 items (Wikidata:WikiProject Top 1000 articles), it is a beginning. User Jeblad has the first 3,130 items. We could use Q1 to Q25000. Regards --Chris.urs-o (talk) 15:48, 3 January 2015 (UTC)

Weird way to choose items. Maybe you can look at the number of bytes/incoming links? Sjoerd de Bruin (talk) 15:50, 3 January 2015 (UTC)

I think I found the logic. A not-confirmed user changes the string of the English label of Qid < 1,000,000 while the string occurs x-times (20, 30, 50) in the item. Opinions? Matěj Suchánek (talk) 14:38, 4 February 2015 (UTC)

Seems ok for me. --Chris.urs-o (talk) 15:49, 4 February 2015 (UTC)

The initial version is Special:AbuseFilter/62. Matěj Suchánek (talk) 17:16, 4 February 2015 (UTC)

This section was archived on a request by: Matěj Suchánek (talk) 09:05, 2 January 2017 (UTC)

Website username

•  Done A filter that warns the user when website username or ID (P554) is being added as reference. A lot of mistakes with that one, it needs to be a qualifier.Sjoerd de Bruin (talk) 07:57, 4 February 2015 (UTC)

Discussion

@Sjoerddebruin: Example? Matěj Suchánek (talk) 14:24, 4 February 2015 (UTC)

@Matěj Suchánek: 1, 2, 3, 4 and more in my recent contributions. Sjoerd de Bruin (talk) 15:39, 4 February 2015 (UTC)

I have created a more general filter (not only P554). Matěj Suchánek (talk) 17:16, 4 February 2015 (UTC)

Thanks! Sjoerd de Bruin (talk) 19:39, 4 February 2015 (UTC)

This section was archived on a request by: Matěj Suchánek (talk) 09:05, 2 January 2017 (UTC)

Description identical to label

 Done I often see new items with the description identical to label, mostly templates. Any idea how to avoid this? Sjoerd de Bruin (talk) 12:55, 11 February 2015 (UTC)

Discussion

@Sjoerddebruin: Unfortunately, the software doesn't provide any information about a present term, only the added one. Instead we can create a more general filter 'bad description' with the logic that descriptions should be different from anything inside item (labels, sitelinks). Is this satisfactory? Matěj Suchánek (talk) 17:51, 11 February 2015 (UTC)

I'm also speaking about avoiding, not to clean up the current mess. We could give it a try. Sjoerd de Bruin (talk) 17:54, 11 February 2015 (UTC)

Indeed, edit filters cannot do anything about the current state. Now you can watch Special:AbuseFilter/64. Matěj Suchánek (talk) 18:08, 11 February 2015 (UTC)

How is that when two languages share the same description? --Pasleim (talk) 18:24, 11 February 2015 (UTC)

Let's wait for some matches but what you suggest, could be problem. Matěj Suchánek (talk) 19:51, 11 February 2015 (UTC)

The filter doesn't seem to work, this and this edit should get caught by the filter. --Pasleim (talk) 10:31, 12 February 2015 (UTC)

I have simplified the filter, so it may catch some other things but it should definitely work now. Matěj Suchánek (talk) 14:44, 12 February 2015 (UTC)

This section was archived on a request by: Matěj Suchánek (talk) 09:05, 2 January 2017 (UTC)

Removal of sex or gender (P21)

 Done Could a filter be created that adds a tag to edits where people remove sex or gender (P21)? There is a lot of vandalism with that, like here, here and here. Sjoerd de Bruin (talk) 07:57, 21 April 2015 (UTC)

Special:AbuseFilter/67 and tag "removal of sex or gender (P21) property". Matěj Suchánek (talk) 18:13, 29 April 2015 (UTC)

This section was archived on a request by: Matěj Suchánek (talk) 09:05, 2 January 2017 (UTC)

Possible advertising

 Done After deleting some pages about non-notable companies, I'm starting to see a pattern in them. There must been a guide on the internet somewhere that tells people how to get in Wikidata as a company. Most items contain properties that are not widely used, like official blog URL (P1581). They also don't contain sitelinks. Sysops can see a example here. Is it possibly to target this with a abuse filter and marking the edits accordingly? Sjoerd de Bruin (talk) 19:13, 10 June 2015 (UTC)

Special:AbuseFilter/70 Matěj Suchánek (talk) 15:52, 11 June 2015 (UTC)

This section was archived on a request by: Matěj Suchánek (talk) 09:05, 2 January 2017 (UTC)

suggestions

 Done Could someone with the authority and ability please modify Special:AbuseFilter/11 to catch these bad words:

• wikt:mierda <-- we get quite a lot of this, e.g. [1]
• wikt:popo e.g. [2]
• wikt:penes (we already have pene, but the filter does not catch penes)
• wikt:tonto
• tu madre (if it's possible to catch a pair of words) "your mother", not exactly bad words, but a very popular expression.

Thanks --Haplology (talk) 06:42, 22 January 2015 (UTC)

@Haplology: Special:AbuseFilter/history/11/diff/prev/691. The filters use regular expressions, so such suggestions are no problem. I don't know Spanish declension, so you may provide more forms of the words to make the filter more efficient. Matěj Suchánek (talk) 14:53, 22 January 2015 (UTC)

Thanks again. The diff is actually hidden from me, but I believe you. I haven't seen any more mierda or penes for a while so it seems to be working. There are also more people watching recent changes. --Haplology (talk) 08:06, 4 February 2015 (UTC)

This section was archived on a request by: Matěj Suchánek (talk) 09:05, 2 January 2017 (UTC)

Badges

 Done Such edits should be catched by a filter. --Pasleim (talk) 09:41, 7 February 2015 (UTC)

one more [3] --Pasleim (talk) 18:14, 7 February 2015 (UTC)

Special:AbuseFilter/history/52/diff/prev/707. Matěj Suchánek (talk) 17:18, 9 February 2015 (UTC)

This section was archived on a request by: Matěj Suchánek (talk) 09:05, 2 January 2017 (UTC)

[[File:

 Done We should completly block the possibility to add [[File: in labels. If the WD-label gets displayed in a WP-article this will cause that a commons file is loaded in its full size and the whole article will be screw up. --Pasleim (talk) 10:36, 3 March 2015 (UTC)

@Pasleim: Special:AbuseFilter/65. I will also create another filter which will log some more wikisyntax. Matěj Suchánek (talk) 18:16, 7 March 2015 (UTC)

This section was archived on a request by: Matěj Suchánek (talk) 09:05, 2 January 2017 (UTC)

A Wikipedia page in need of a description.

 Done Through the new mobile app, people are adding the description A Wikipedia page in need of a description., see [4], [5], [6]. There should be a filter which blocks this text. Thanks. --Pasleim (talk) 13:22, 22 July 2015 (UTC)

Special:AbuseFilter/75. Matěj Suchánek (talk) 13:49, 22 July 2015 (UTC)

This section was archived on a request by: Matěj Suchánek (talk) 09:05, 2 January 2017 (UTC)

Prevent adding links to new user pages for new users

Many spammers create user pages with weblinks and this is their first edit. I think we can set an abuse filter to disable creating user pages for user with 0 edits and very young (<1d?)--GZWDer (talk) 11:08, 4 February 2015 (UTC)

Something is in Special:AbuseFilter/4 but you can't see it. Matěj Suchánek (talk) 14:22, 4 February 2015 (UTC)

This section was archived on a request by: Sjoerd de Bruin (talk) 10:19, 28 September 2017 (UTC)