User talk:Matěj Suchánek

Stop importing labels for proper names

Please don't import redundant labels from Wikipedia [1] Use [mul] instead for names. Matthias M. (talk) 18:03, 22 July 2025 (UTC)

My bot already respects mul whenever it's present. The more mul labels are added, the less labels will my bot import. Note that some users still dispute whether mul is appropriate for (all) names. --Matěj Suchánek (talk) 12:36, 23 July 2025 (UTC)

No, it reimports English labels when [mul] is present and [en] was already cleaned up. Special:Diff/2410577900 This is not helpful. Matthias M. (talk) 19:27, 1 October 2025 (UTC)

@Matthias M.: I see. But presence of mul, in my opinion, should not be an ultimate blocker for importing other labels. (Think of languages like Arabic.)

Also, this specific case is weird. The English page has only a slightly different title:

• en: Burnout Paradise: The Ultimate Box
• mul: Burnout: Paradise - The Ultimate Box

Why isn't exactly the same? --Matěj Suchánek (talk) 13:00, 25 October 2025 (UTC)

If you are searching for a rule to avoid misbehavior of your bot, then don't import when the new label is already a synonym for [mul]? Matthias M. (talk) 13:16, 25 October 2025 (UTC)

Good idea. Also when it's already a synonym in the same language (provided there is already a [mul] label). --Matěj Suchánek (talk) 19:12, 26 October 2025 (UTC)

Incorrect vandalism report for Q96955587 by MatSuBot

Hello,

This item (Mathieu Roy (Q96955587)) contained confusing information due to a homonymy. It was mixing data for the Canadian visual artist Mathieu Roy (Mathieu Roy (Q96955587)) with a homonym (a French researcher).

A cleanup was performed to remove the information and identifiers (like SUDOC or VIAF) that actually belonged to the French researcher, in order to resolve the confusion.

Thank you. --Mateo.Mfalme (talk) 11:05, 25 October 2025 (UTC)

Hello, not everything on User:MatSuBot/Reports/Vandalism 20251013 is necessarily a vandalism (if I were sure it was, I would have the bot undo it right away), it's just suspected vandalism based on some heuristics. In this case, it was triggered by these three removed/overwritten [2][3][4] statements with proper references. --Matěj Suchánek (talk) 13:07, 25 October 2025 (UTC)

Hello Matěj,

Thank you for the clarification regarding the bot report.

I can confirm that this was indeed a false positive. As I explained in my initial message, the item Mathieu Roy (Q96955587) was confusing two different people due to homonymy (a Canadian artist and a French researcher).

The edits that triggered the suspicion—the removal of the three referenced statements [3][4][5]—were necessary because those statements and their identifiers belonged to the Canadian artist, not the French Resaercher this item is about.

I understand why removing referenced statements would look suspicious to the bot. Perhaps this incident provides a good opportunity to help refine the heuristics to better distinguish necessary cleanup of homonymy from actual vandalism.

Thanks again,

--Mateo.Mfalme (talk) 13:41, 25 October 2025 (UTC)

Znicenekostely

Ahoj, bylo by možné u MnM katalogu Znicenekostely změnit formatter ID položky katalogu z http://www.znicenekostely.cz/?load=detail&id=$1 na (aktuálně funkční) https://www.znicenekostely.cz/objekt/detail/$1 ? Jklamo (talk) 15:52, 30 November 2025 (UTC)

Ahoj, já tohle oprávnění asi nemám. Ani menu https://mix-n-match.toolforge.org/#/import/1545 mi to nenabízí. --Matěj Suchánek (talk) 16:08, 30 November 2025 (UTC)

Special:Diff/2442489875

Radšej nemeň sk a en názov, lebo existujú 2 rezervácie, ktoré majú identický názov (a teda rozlišujeme poznámkou v zátvorke), ale sú pri tom iného charakteru. Viac viď úvod na skwiki stránke. Dušan Kreheľ (talk) 14:39, 16 December 2025 (UTC), --Dušan Kreheľ (talk) 17:25, 16 December 2025 (UTC)

Ta věc ale se jmenuje "Fintické svahy", ne "Fintické svahy (prírodná rezervácia)". Jestli je to přírodní rezervace nebo evropsky významná lokalita, o tom informuje popisek. --Matěj Suchánek (talk) 14:54, 30 December 2025 (UTC)

@Matěj Suchánek: Tak to daj ako diskusiu. … Proste, vyzerá to tak, že v praxi názov položky nemusí byť podľa oficiálneho názvu, viď Q20860409 alebo názvy štátov nap. Q214 so skrátenou formou. Dušan Kreheľ (talk) 15:11, 30 December 2025 (UTC)

Nikde jsem nezmínil koncept oficiálního názvu. Štítek je jednoduše „to, jak se tomu říká“. „Fintické svahy (prírodná rezervácia)“ tím nenazývá nikdo. Kromě Wikipedie, která je omezená unikátními názvy stránek. Ale Wikidata tohle omezení naštěstí nemají, tak není potřeba tento workaround přebírat.

Nemluvě o dalším nesmyslu, kterým je „Fintické svahy (europe nature protected area)“ jako výchozí hodnota pro všechny jazyky. Jak může být „europe nature protected area“ něčím, o čem se dá očekávat, že je společné velkému množství jazyků? Matěj Suchánek (talk) 15:49, 30 December 2025 (UTC)

@Matěj Suchánek: Ok, obe Fintické svahy sú wikidata názvom iba ako Fintické svahy a v aliasoch som to upresnil. Takže obe by už mali byť okey. Dušan Kreheľ (talk) 16:34, 30 December 2025 (UTC)

Request for verification and validation of translation markup

Hello @Matěj Suchánek, Thank you very much for taking the time to review and correct the translation markup of the page Wikidata:ArchiveExternalLinks. I truly appreciate your support and attention to detail. Following these corrections, I would like to kindly ask whether you could verify and validate the translation markup, so that I may proceed with translating the page into multiple languages. Thank you again for your valuable assistance. I look forward to your feedback. Poro26 (talk) 16:16, 10 January 2026 (UTC)

The page is now ready for translation. --Matěj Suchánek (talk) 16:34, 10 January 2026 (UTC)

Q2587336

Hello, the proper version to restore is this one. The one you've chosen misses some facts, e.g. date of birth. Komarof (talk) 18:14, 7 March 2026 (UTC)

Hello, thanks for letting me know. I have undone the initial series of disruptive edits. Feel free to improve the item yourself, the lock is not exclusive for admins. --Matěj Suchánek (talk) 18:26, 7 March 2026 (UTC)

Voice Acting (Q22920017)

Hello, can you add the the portuguese page in the languages links?

https://pt.wikipedia.org/wiki/Voz_Original Thagner Moss (talk) 20:24, 15 March 2026 (UTC)

 Done --Matěj Suchánek (talk) 20:28, 15 March 2026 (UTC)

Edit of you bot (Q164597)

Hello, I want to ask if this edit of your bot was something that is regular on Wikidata, i.e. copy-pasting an entry's Wikipedia title as its label. Michael21107 (talk) 04:15, 21 March 2026 (UTC)

No, this is a backup mechanism for when the user doesn't insert the label themself. There is usually at least a week before it happens.

If you are referring to the parenthesized appendix, that's obviously incorrect, but unfortunately not always, so the robot tries to guess (for example, based on the description or the other labels). --Matěj Suchánek (talk) 17:47, 23 March 2026 (UTC)

Should every item with a linked Wikipedia have a label for that language, and regardless, is there an official policy on that? Michael21107 (talk) 02:56, 18 April 2026 (UTC)

The question is weird to me. Ideally, every item should have a label (or WD:MUL label) and description in every language. This is a never-ending effort, and we can some manual effort by copying the name of the Wikipedia page in the corresponding language, because in the vast majority of cases the wanted label is same. --Matěj Suchánek (talk) 17:05, 26 April 2026 (UTC)

Vandalism bot questions!

Hi Matěj, francisco here. I am a product manager on wikimedia enterprise. I'm wondering about your vandalism report bot - https://www.wikidata.org/wiki/User:MatSuBot/Reports/Vandalism_20240318 -

Do you have documentation you can share with me about how it works/what it captures? I wondering if it can be a useful and dependable bit of metadata to include in our APIs.

thanks for any info. also feel free to reach me at fnavas@wikimedia.org FNavas-WMF (talk) 18:29, 25 March 2026 (UTC)

Hello, Francisco! I am glad you reached me.

I have described the idea in this post: Special:PermaLink/2128794327#On Vandalism Tracking.

Basically, there is a system called Property Constraints. It was developed by the community quite soon after Wikidata was created (and later it was integrated into the Wikidata interface). The constraints are "rules" about what information a property call hold and is sensible. This set of rules is curated by the community, and it's a means to control the otherwise unrestricted data model of Wikidata (i.e., you can enter basically anything as someone's mother, but only women born before them are sensible). The most notable application of the system has been the regular compilation of lists of items that violate each of these constraints (for example, Wikidata:Database reports/Constraint violations/P6 for head of government (P6)) by a bot.

Over time, I noticed that some vandalism made to properties can be identified by looking up if and how many constraints an edit violates (unfortunately, this information is not provided by the interface). So I decided to create a Pywikibot-based Python library that implements:

• most of the constraint types (focusing on queries formulated as "which constraints are violated in this revision" instead of "give me the list of items which now violate this constraint") and some custom rules,
• interpretation of Wikidata edits (query: "what was changed by this edit(s)"),
• an efficient lookup of constraints a particular edit needs to be tested for (so that the bot does not have to load all of them),
• an improvized ranking of severity.

Every week, I have the bot run over unpatrolled recent changes (i.e., the last month), check every edit (or a "streak" thereof, which is more accurate) for newly violated constraints and report them in a table (one weekly report takes around 2 hours).

There is still a plenty of things to address that I haven't got to. For example, I haven't involved machine learning so far, which could perhaps improve the ranking or make this system suitable for autonomous real-time patrolling (I'm actually doing that, too, but only for a fixed set of rules).

As for what it would take to have this implemented server-side, the most critical missing piece is that the action=wbcheckconstraints API module considers only the most recent revision of an item and cannot evaluate an older one. If this was supported, then you would just query for the current and previous revision and observed the difference. (But this would still be somewhat inefficient.)

--Matěj Suchánek (talk) 10:08, 26 March 2026 (UTC)

Creating Mix-n-match catalog

What are the advantages and disadvantages of scraping from URL vs uploading a file? Which one should i use? Trade (talk) 02:03, 30 March 2026 (UTC)

Depends what it takes to obtain a file. If you can obtain a file repeatedly (for catalogs that are never complete) without scraping, it's a no-brainer. You also reduce the risk of the scraper getting banned (which is nowadays more likely than before due to AI agents' traffic).

However, if you can obtain the file only by scraping, then you should just hand the labor over to a scraper, which will regularly do the same for you. --Matěj Suchánek (talk) 06:51, 30 March 2026 (UTC)

Problem is that Preliminarily matches and manually sync catalog doesn't seem to work with any catalog i make Trade (talk) 14:55, 30 March 2026 (UTC)

MatSuBot

Hello. What happened here? Eurohunter (talk) 18:43, 7 April 2026 (UTC)

Hello. It's because pl:Wojna dronowa → d:Q30588142#sitelinks-wikipedia. --Matěj Suchánek (talk) 06:10, 8 April 2026 (UTC)

Please stop duplicating [mul]

when adding labels from sitelinks. Make your check case insensitive. Matthias M. (talk) 21:03, 9 April 2026 (UTC)

Can you create cross-property moveclaim?

I need a moveClaim analog which allow 'URL to string' claim copy to insert KCI article ID (P14184) data. How we can do it? Toolipo (talk) 16:03, 10 April 2026 (UTC)

Additionally, it would be ideal if there were a function to copy only the first or last n characters of the string. Toolipo (talk) 16:07, 10 April 2026 (UTC)

You can set up a (temporary) {{Autofix}} that KrBot periodically performs. --Matěj Suchánek (talk) 16:23, 10 April 2026 (UTC)

Q97323205

Dear Sir,

The page was ours and was hijacked by Drive Chauffeur. We contacted them and no action. We were disappointed as we had to edit all our schema. I hope you can consider please. Thank you. ~2026-24256-50 (talk) 22:31, 20 April 2026 (UTC)

Thanks for the undos of merges

Thanks very much for your undos (and also for adding different from (P1889)! As I have written in the explanation of the QS batches, these are based on the match of VIAF or ISNI, so they can contain mistakes (I estimate a low percentage, somewhere around 1%-2%). You can always undo my edits as you did, so that I notice the issue and I go on the items and fix the VIAF/ISNI; in the three cases you undid today, the VIAF was wrong on the older item in all cases, now they are fixed. Best, --Epìdosis 14:33, 26 April 2026 (UTC)

https://www.wikidata.org/wiki/Q97323205

Hi Matej, Being one of the active admin, there are a lot of tasks on your plate. Your efforts to have a quality Wikidata proves you are truly dedicated to make things better. The above listing was hijacked by Drive Chauffeur and we cant help but to be disappointed. I wonder if there's a way for us to update it. Thank you.

Dinez Carnay Founder Dinez Taxis and Airport Transfers ~2026-24256-50 (talk) 13:19, 19 May 2026 (UTC)

Thank you Matej for your kind understanding of the circumstance and for exerting efforts in alignment. Sincerely yours, Dinez Carnay ~2026-24256-50 (talk) 21:41, 23 May 2026 (UTC)

MatSuBot reverted my edit in Q3342 (Côte-d’Or)

Hello. Why does it happen and how can it be stopped? The term I added is the disambiguating term "מחוז" (a district), that will be used in the name of the article in Hebrew when it is written. Until then, removing the disambiguating term causes item Q3342 to link to the disambiguation page in one of our queries. Regards, Virant (talk) 16:11, 4 June 2026 (UTC)

The thing that the bot edited is label. Labels link nowhere. See also Help:Label#Disambiguation information belongs in the description. --Matěj Suchánek (talk) 17:13, 4 June 2026 (UTC)

True, labels link nowhere, but when our bot is retrieving missing articles in Hebrew, it is suggesting an optional linked name based on this label. So we check and update a disambiguation value to such labels (after considering existing and desired article names), update the Wikidata label accordingly, it all works well after out next bot run, and then MatSuBot goes and deletes it...

I read what you sent, and I agree with it When the article exists, which is not the case here. Can you please exclude this deletion in cases where there is no matching link to an article? (if "hewiki" value doesn't exit, don't change "label / he". I guess such exclusion will be useful for any language). Thanks, Virant (talk) 02:18, 23 June 2026 (UTC)

@Virant: If I understand this right, you are using the label as a placeholder name for the corresponding "red link", because Wikidata does not support red links (phab:T212211).

I am repeating myself, but this is not a good idea. Wikidata labels should be useful regardless of whether a Wikipedia article is present. Wikimedia is not just (Hebrew) Wikipedia, and Wikidata is not used only by Wikimedia. I totally understand doing this via Wikidata is very convenient, but I recommend not using the label for this.

Of course, modifying my robot with a simple rule like that is trivial. But what if someone else appears to do this clean up? They are not doing anything wrong, only following what is written somewhere else. Would there be an edit war? Who is right, then? --Matěj Suchánek (talk) 09:04, 11 July 2026 (UTC)

Sonny Liston's date of birth

Good day,

There are many sources that dispute Sonny Liston's date of birth as 8 May 1932. There is an entire section with sources discussing it on his Wikipedia page.

https://en.wikipedia.org/wiki/Sonny_Liston#Date_of_birth

As you can see, Liston only believed that his date of birth was 8 May 1932 and used that date for official purposes, however, there is no official record of it. He is absent from the 1930 census and is listed as being 10 years old in the 1940 census. The state of Arkansas, where he was born, did not make birth certificates mandatory until 1965. DeathTrain (talk) 11:59, 21 June 2026 (UTC)

No response? DeathTrain (talk) 14:04, 27 June 2026 (UTC)

Sorry, busy week.

The point of Wikidata is to collect what is stated in sources. So if there are 6 sources claiming "Birth date: 1932-05-08", you cannot simply disregard it.

It's quite common that sources disagree or are uncertain, that's why Wikidata does not restrict inserting only one birth date per person, and allows ranking claims or specifying uncertainty.

So what is the ultimate value that Wikidata should present most prominent? --Matěj Suchánek (talk) 07:25, 28 June 2026 (UTC)

Probably certainty. So long as there are several credible sources disagreeing with that birth date and specifically saying that his date of birth is uncertain, you cannot simply disregard those either. Just because one date is sourced more, that does not make it more true, especially in this context. Concerning ranking claims, I do not think that it is very appropriate in this situation as this as this data is often used on Wikipedia (such as on French Wikipedia: https://fr.wikipedia.org/wiki/Sonny_Liston), and by presenting one claim as ranked higher than the other, I feel that is violating https://en.wikipedia.org/wiki/Wikipedia:Neutral_point_of_view. Given how uncertain his date of birth is, I think the most neutral thing that can be done is to not state any date of birth at all. I do not think it is appropriate to put in any date, when it is highly likely to be wrong. I am even reluctant to put in the date of July 22, 1930, which was what was proposed in this source: https://tss.ib.tv/boxing/a-birthday-for-sonny-liston/ DeathTrain (talk) 01:27, 1 July 2026 (UTC)

Any reply? What I did was I put that year of 1932 is presumed, but not the specific date of May 8, 1932. I personally still prefer no date at all, given how uncertain his date of birth is, and I will probably remove even that unless you or someone else convinces me otherwise. DeathTrain (talk) 00:21, 11 July 2026 (UTC)

I am even reluctant to put in the date of July 22, 1930, which was what was proposed in this source[.] There is no reason for being reluctant. Wikidata stores verifiable claims, not "facts" (WD:Introduction, Help:Statements#Add only verifiable information, Help:Statements#Plurality and consensus). We will probably never know the truth, but we can (and should!) record all the claims, so that everyone has enough information to make a decision. Disregarding relevant information, especially when it's widespread, is what I would call a violation of NPOV. (It's a matter of time someone or some robot adds it back without providing the context.)

I think the most neutral thing that can be done is to not state any date of birth at all. This contradicts nothing of the above. "1930s" is valid claim that can sit next to all the other claims, while being the preferred value. --Matěj Suchánek (talk) 09:18, 11 July 2026 (UTC)

Stop adding nonsense "titles"!

Please stop adding nonsense "titles" like here. "Wikipedia:Redaktion Film und Fernsehen/Glashütte/Insidious 6" is NOT the title of the movie. It is the title of a Wikipedia article that has not even been released and should not be at Wikidata at all. The "title" already had been removed 3 months ago but what is the point of that if this nonsense is always being readded. ~2026-38584-26 (talk) 11:43, 7 July 2026 (UTC)

Indeed, these pages should not be connected to Wikidata at first. I have disconnected all of them, fixed labels where needed, and made the bot not import these pseudotitles anymore. --Matěj Suchánek (talk) 15:00, 11 July 2026 (UTC)

Invitation: Users with Extended Rights conversations

Hi Matěj,

I am not sure if you have seen it already but we are hosting sessions to better understand the workflows and issues of users with extended rights on Wikidata and it'd be great if you'd join. More details at Wikidata:Users with Extended Rights conversations 2026.

Cheers Lydia Pintscher (WMDE) (talk) 13:36, 28 July 2026 (UTC)

Request for Confirmed rights for MichealBVN

Hello. I have completed 4 days and 65 edits, but my account has not been automatically confirmed yet. Could you please grant me the 'Confirmed' rights manually? Thank you! -- MichealBVN (talk) 17:35, 16 August 2026 (UTC)

Pywikibot will drop support for Python 3.9

Hi Matěj,

I would like to inform you that the next major release, Pywikibot 12, will drop support for Python 3.9. Python 3.9 will remain supported in Pywikibot 11.X. Development of Pywikibot 12 is scheduled to start in late October or early November, and the stable release will be published no earlier than one month later. If there are any reasons to postpone this schedule, please bring them up at phab:T423572.

Best,  @xqt 18:12, 28 September 2026 (UTC)

Abuse filter 132: new item for PublicAML blocked

Hello Matěj. I am a new user and tried to create an item for PublicAML, a free cryptocurrency anti-money laundering screening service (official website: publicaml.org). Every attempt, even with only a label and a description, was disallowed by abuse filter 132 ("specific spam items"), which I understand you maintain.

I am the founder, so I have a conflict of interest; it is disclosed on my user page. The item would be factual and sourced: official website, inception date (10 September 2025, launch announcement), X / LinkedIn / GitHub / Telegram identifiers, the npm package @publicaml/mcp-server and its source repository. The service is also listed in the official MCP server registry.

Could you either create the item or let me know what is needed so I can create it myself? I am not trying to work around the filter. Thank you. Saloprj (talk) 07:53, 29 September 2026 (UTC)

Abuse filter 132 false positive: new item for PublicAML

Hello. I tried to create an item for PublicAML (https://publicaml.org), a free cryptocurrency anti-money laundering screening service, and every attempt was blocked by abuse filter 132 ("specific spam items") — even with only a label and a description.

I am the founder, so I have a conflict of interest; it is disclosed on my user page. The item would be factual and sourced: official website, inception date (10 September 2025, launch announcement), X / LinkedIn / GitHub / Telegram identifiers, the npm package @publicaml/mcp-server and its source repository. It is also listed in the official MCP server registry.

Could an administrator either create the item or let me know what is needed so I can create it myself? Thank you. Saloprj (talk) 07:54, 29 September 2026 (UTC)

Sorry, this is an accidental duplicate of my request in the section above — please reply there. Saloprj (talk) 07:55, 29 September 2026 (UTC)