User talk:Mike Peel/Archive 2

From Wikidata
Jump to navigation Jump to search

Culture of Pescara

Dear Mike, for sure it was a distraction, but an item for "Culture of Pescara" already existed on 'data, Q26204988. Now I'm going to merge the two items. Regards, -- Blackcat (talk) 07:43, 8 July 2019 (UTC)

@Blackcat: Thanks for merging the two! I have some code now that uses the commons sitelinks to try to match up new enwp pages, but that didn't work here as there wasn't a commons link (and the search also didn't work due to the language difference). So I fell back to creating a new page instead. Hopefully that should become less common over time, as we get more commons sitelinks. :-) Thanks. Mike Peel (talk) 07:52, 8 July 2019 (UTC)

Emijrpbot thanks

Belated thanks for taking over Emijrpbot 6, which I see you've now put on an hourly schedule. I keep seeing Pi bot doing excellent things helpful to Women in Red & each time take my hat off to you.

Meanwhile here, btw, is another list of en wiki articles with no wikidata. It's just been emptied in the last day or so (stats) by GWZflood, which creates lots of propertyless items. Not sure if it's any use to you in locating new candidate articles? I think Magnus updates it once daily. --Tagishsimon (talk) 09:39, 8 July 2019 (UTC)

@Tagishsimon: I think in principle it can be run on different input sets. However, I've stopped the bot for now due to the issue in the section directly above. I'll try to do some reworking of it later this week (but it's going to be a busy week!). Thanks. Mike Peel (talk) 09:44, 8 July 2019 (UTC)
Noted; sorry if I've brought ropey code to your door. I hope you get time to fix it, but equally hope you take your own good time to do so. Nothing's burning. --Tagishsimon (talk) 12:42, 8 July 2019 (UTC)
@Tagishsimon: It's nice code, this is just a case of expanding its capabilities. :-) It's now running again in test mode, let's see how that goes. Thanks. Mike Peel (talk) 20:57, 12 July 2019 (UTC)

@Tagishsimon, Bahnmoeller, M2k~dewiki: I've returned the bot to normal service now, please let me know if you spot any further issues. It also now looks through the Duplicity lists (en, fr, de) once per day. Thanks. Mike Peel (talk) 11:04, 13 July 2019 (UTC)

@Tagishsimon, Bahnmoeller, M2k~dewiki: Checking in, has everything looked OK for this last week? Thanks. Mike Peel (talk) 17:20, 19 July 2019 (UTC)
Hello Mike, from my point of view it looks OK so far. When you check the history of Pi bot you might find some duplicate objects, which the bot created and already have been merged (for example d:Q65659760, d:Q65659053, d:Q65659049, d:Q65657800, d:Q65657766, d:Q65643848, d:Q65621776, d:Q65620600, d:Q65600533, d:Q65589499, d:Q65589029, d:Q65562662, d:Q65562555, d:Q65560840, d:Q65560670, d:Q65560552, d:Q65559810, d:Q65552898, d:Q65551320, d:Q65550120, d:Q65112883, d:Q65112875, d:Q65098247, d:Q65095981, d:Q65096270, d:Q65096268, ...). I guess, it would be hard to avoid duplicates in some (special) cases, but maybe you can improve the code, based on the examples. --M2k~dewiki (talk) 17:58, 19 July 2019 (UTC)
Another thing that I have observed: it seems, that the boot does not create objects when the article is moved from the user namespace to the article namespace, it only creates objects for articles directly created in the article namespace (?!?). --M2k~dewiki (talk) 18:26, 19 July 2019 (UTC)
@M2k~dewiki: Thanks for the feedback. Most cases of the duplicates are where the name of the article isn't in the Wikidata labels, it's difficult to avoid that by bot. The bot looks for new pages, which doesn't include the ones moved from user space. I tried a test run today looking through unconnected pages, but that needs some more work to only look at mainspace articles (it created a few items for category items and others). I'll have a think. Thanks. Mike Peel (talk) 16:03, 20 July 2019 (UTC)

Bring back control

It's depressing to see anti-Wikidata propaganda follow Brexit tropes. (Although the issue is a rather real and simple one i.e. the poor state of watchlisting and lack of direct links to Wikidata edit buttons.) Nemo 17:04, 22 July 2019 (UTC)

@Nemo_bis: I just tune out when it becomes an "us vs. them" argument rather than "how do we improve this together". Thanks. Mike Peel (talk) 18:01, 22 July 2019 (UTC)

Pi bot bad edit

Please check how this edit by pi bot came about. https://www.wikidata.org/w/index.php?title=Q55106039&type=revision&diff=987514129&oldid=987503267  – The preceding unsigned comment was added by Masegand (talk • contribs).

@Masegand: It was because that's what's stated in the article - {{death date and age|df=yes|0|0|0|0|0|0}} is there, albeit hidden in comments (which the bot doesn't notice). I've added an exception for birth/death dates of 0-0-0, hopefully that should avoid this recurring. Thanks. Mike Peel (talk) 09:11, 28 July 2019 (UTC)
Ok thanks. But it would be better if the bot would not act on commented sections as if it was not commented?--Masegand (talk) 09:18, 28 July 2019 (UTC)
Probably, but that would be much more complicated. The code is open source, if you want to suggest a modification to it, though! Thanks. Mike Peel (talk) 09:21, 28 July 2019 (UTC)

Golden Ballet

Hi! For information, Pi bot considered Golden Ballet (Q66730466) as a human instead of a racehorse: [1]. Regards, Korg (talk) 00:16, 25 August 2019 (UTC)

@Korg: They're the same thing, just with different numbers of legs, right? ;-) I've tweaked the code so that it avoids articles that contain the word 'racehorse', so hopefully the mistake shouldn't happen again. Thanks. Mike Peel (talk) 07:57, 25 August 2019 (UTC)

Q29784568

Hello, there is a little problem in the Wikidata-Object Q29784568, which I don't know how to solve. The box included an image-file, which showed a bridge of a different location (en:File:Coltura Brücke.JPG); in the last edit I changed the file to the correct one (en:File:Ponte di Stampa - panoramio.jpg), however on the commons-page en:Category:Ponte sulla Maira, Stampa there is still the old picture shown in the Wikidata Infobox. Could you please help? Thanks and greetings --Muck50 (talk) 09:55, 28 September 2019 (UTC)

@Muck50: Thanks for correcting the picture! It looks OK to me in the Commons category, so I think it's a caching issue. Can you try visiting [2] to purge the cache, and see if that looks OK to you? I've also made a null edit (clicking edit, then saving the page without changing anything), which might also help. Thanks. Mike Peel (talk) 09:58, 28 September 2019 (UTC)
Thanks for your fast reply! Everything is fine now! --Muck50 (talk) 10:00, 28 September 2019 (UTC)

New page for catalogues

Hi, I created a new page where I started collecting sites that could be added to Mix'n'match and I plan to expand it with the ones that already have scrapers by category. Feel free to use, expand. Best, --Adam Harangozó (talk) 10:01, 17 October 2019 (UTC)

Pi bot bug?

Hi Mike, I've just spotted an interesting bug here - it imported a date of death from enwiki because there was a {{Death date and age|...}} template, but it was commented out rather than intended to display. (Not sure why, probably copied over from somewhere - there is also a dummy {{Birth date|...}} template in comments). Do you know if it would be possible for the bot to look out for these, or is it such a rare problem it's not worth worrying about? Andrew Gray (talk) 19:28, 24 October 2019 (UTC)

@Andrew Gray: I don't think it's trivial to figure out whether the template is actually used on the page or not. The bot reads the source code only, not the rendered page, so it would need to try to figure out what is enclosed in the comments and what isn't... Thanks. Mike Peel (talk) 07:00, 25 October 2019 (UTC)

Czechoslovakia

Ehm Mike, I somewhat suspect that your bot isn't that skilled in geography. Indeed he has just assigned to Czechoslovakia the troops of His Majesty the late George VI of the United Kingdom -- Blackcat (talk) 18:33, 25 October 2019 (UTC)

@Blackcat: That's not my bot's fault - blame @JAnDbot: back in 2016! [3] Thanks for cleaning it up! Mike Peel (talk) 18:53, 25 October 2019 (UTC)

PI bot a bit overeager

Hi there! Never had an issue with your bot before, but recently I've seen it be a bit "over eager" adding gender, occupation, birth date etc. to items that aren't humans. Here's a fresh example: https://www.wikidata.org/w/index.php?title=Q72328954&diff=1038891758&oldid=prev Could you have a look when you get the time? Moebeus (talk) 00:19, 26 October 2019 (UTC)

@Moebeus: According to the categories at en:Saigon discography, Mr Discography is a living African-American rapper born in 1977. I'd suggest cleaning up the categories there... Thanks. Mike Peel (talk) 05:54, 26 October 2019 (UTC)
@Moebeus: Thinking about it, what I can do is add an exception to avoid articles with 'discography' in the titles. I can't code that up today, as I'm travelling, but I can look at it next week. Are there any other cases/patterns you've spotted that the bot should avoid? Thanks. Mike Peel (talk) 07:07, 26 October 2019 (UTC)
@Mike Peel: "Mr. Discography" hehe. As I said, I've never really had an issue with the bot before, but I started noticing it for the first time here: https://www.wikidata.org/w/index.php?title=Q72088490&action=history Looking at the wikipedia article it's easy to see why the bot is confused, but looking at the WD item it really shouldn't be any doubt that the item is a group (of humans) and not a human. Thank you for taking the time and let me know if there's anything I can do to help (bar messing with wikipedia categories, not touching that 😄) Moebeus (talk) 09:47, 26 October 2019 (UTC)

Bot does not work properly

I have reverted this because it is nonsense. -- MovieFex (talk) 12:35, 9 November 2019 (UTC)

@MovieFex: Thanks for reverting it. It looks like it came through the persondata at the bottom of this version - garbage in, garbage out (Q1569381). Thanks. Mike Peel (talk) 12:45, 9 November 2019 (UTC)

AfroAmerican engeneer

Hoi, that category does not distinguish people for their speciality. It is why I removed a bot entry at Achille Messac thanks, GerardM (talk) 11:18, 2 December 2019 (UTC)

Same problem with Pi bot

Hello! I would like to report that the problem with Pi Bot which we had discussed earlier is still persisting. Veverve (talk) 15:39, 8 December 2019 (UTC)

@Veverve: As discussed earlier, this is by deliberate design, it is not a bug. If you want to change how this works, please get community consensus to do so. At the moment, the convention is that the Commons category goes on the Category item where it exists, and there is Lua code that can follow from the topic item to the category item to retrieve the Commons category link from articles. Thanks. Mike Peel (talk) 16:38, 8 December 2019 (UTC)

Please Review this page

Kindly review this page https://www.wikidata.org/wiki/Q77359124 (Shubham Ghodke) 17:55, 12 December 2019 (UTC)

@Shubhamghodke3904: Why? It looks fine to me. Thanks. Mike Peel (talk) 21:25, 12 December 2019 (UTC)

Hello Mike Peel,

I just noticed that Q25548561 contains a link to en:Bilal Nagar although this page on en-wikipedia was deleted on 8th November 2019. I have no idea whether there are more such example or not.Robby (talk) 19:06, 14 December 2019 (UTC)

@Robby: It's probably worth raising a ticket at phab about this. @Lydia Pintscher (WMDE): any ideas? Thanks. Mike Peel (talk) 19:08, 14 December 2019 (UTC)
Phabricator ticket created T240762Robby (talk) 19:37, 14 December 2019 (UTC)

Hubert Jäger

Hello Mike,

Pi-Bot added a false Birth Date at Q76752945. Hubert Jäger was born at 18 March 1959 and not at 18 March 1859 . Your Bot added a birthdate 100 years before the real one. Can you please check why that happened. I have seen that mistake when I read this [4] article about the Wikipediacommunity in Hamburg where Hubert Jäger was mentioned. In Google after searching for Hubert Jäger I saw that the calculated age is 160 years and then looked into Wikidata. I think this is a bad thing, that such information comes to the public and such information should be checked before the Information is online in Google results. Do you have an idea how things like that can be checked. I think this is a more or less easy case because there is no known human older than 122 years. -- Hogü-456 (talk) 14:32, 22 December 2019 (UTC)

@Hogü-456: The 1859 date was entered by @Wikipeter-HH: in the original version of the German Wikipedia article, it was later fixed there, but not here. I'm not sure how Pi bot could automatically check for errors like this - the absence of a death date doesn't necessarily mean that someone is still alive, it could just be that we don't know it. Thanks. Mike Peel (talk) 15:17, 22 December 2019 (UTC)
Apologies for being the cause of confusion. The point here is, that the error happened in the 'Person Data' template, which is only visible in edit mode, not when reading an article. The date given in the article text was correct. One possible check here could be, when there is a person older than, say, 110 years without a death date, to compare the date in the 'Person Data' template with the date given in the article text ... Regards --Wikipeter-HH (talk) 15:58, 22 December 2019 (UTC)
Compare the date in the 'Person Data' template with the date given in the article text is a good idea. Is there a way to do that automatically. I think the number of possible notations of the date is not so high and so that could be done automatically but I am not a programmer. Is it possible to get the text of the first sentence of an article with a query. Then it could be checked if this is the same. I think in a Spreadsheet it were possible to check if the date is the same or not. -- Hogü-456 (talk) 21:40, 22 December 2019 (UTC)

Birthdate Isabel dos Santos

Hi, see the wayback link in the German Wikipedia and add it yourself. It does not fit in the English Wikipedia, who knows why. 179.13.66.167 20:14, 19 January 2020 (UTC) http://multimedia.fnac.pt/multimedia/PT/pdf/9789897412257.pdf

Add it as a reference using reference URL (P854), please. Thanks. Mike Peel (talk) 20:19, 19 January 2020 (UTC)

OK later tomorrow. 179.13.66.167 01:07, 20 January 2020 (UTC)

Hi Mike Peel, I tried it but I was not able to fix that. Could you do that for me please...!! Thanks. 179.13.66.167 15:15, 20 January 2020 (UTC)
Belatedly done, thanks for providing the reference. Thanks. Mike Peel (talk) 04:31, 24 January 2020 (UTC)

"Pi bot" stopped creating WD-objects for persons

Hello Mike, it seems that "Pi bot" stopped creating WD-objects for persons on 20th of January:

@M2k~dewiki: It's running again now after a reboot. I've also managed to get a new set of commons categories to add the infobox to, so that's also now running. Thanks. Mike Peel (talk) 11:11, 25 January 2020 (UTC)

Seychelles sculpture categories

I added the Commons link but removed it when adding the category combines topics (P971) - for one, "sculpture" is art of sculpture (Q11634), for the other, it is sculpture (Q860861). For other countries there are separate items for example Category:British sculpture (Q8315823) (not only specific sculptures, or located in the UK) and Category:Sculptures in the United Kingdom (Q8723820) (located in the UK, but not necessarily British). Are these usually combined in one category? Peter James (talk) 11:09, 25 January 2020 (UTC)

@Peter James: I'm not sure of the ideal way to model it, but since we only have one enwp category and one commons category it makes sense to link them, at least for now. Thanks. Mike Peel (talk) 11:14, 25 January 2020 (UTC)

Interpretations by Pi bot

Hi there,

Can you explain these edits, which Pi bot claims to be based on this version? How comes Pi bot comes up with '2 March 2019'?

Thanks for your answer, RonnieV (talk) 12:21, 3 February 2020 (UTC)

@RonnieV: It's something to do with the malformed value in the infobox, "19 March {{death year and age|1971|1884}}", I'll probably have to code an exception to handle that. I'll try to investigate it this evening. Thanks. Mike Peel (talk) 12:24, 3 February 2020 (UTC)
Thanks for your quick answer. An exception would be nice. Good luck with the investigation. RonnieV (talk) 12:42, 3 February 2020 (UTC)
@RonnieV: OK, the code now supports date year and age, and just extracts the death year from it. Code change at [5], and test edit at [6]. I've updated the main bot code with this version, please let me know if you spot it adding anything else odd! Thanks. Mike Peel (talk) 19:30, 3 February 2020 (UTC)

Congratulations, Dear Administrator!

English | español | français | العربية | Nederlands | русский | +/−

An offering for our new administrator from your comrades... (our gift is better than the one at Commons or Meta)

Here's your badge: {{User admin}}/{{#babel:admin}} and {{Admin topicon}}. Enjoy!

Congratulations! You now have the rights of administrator on Wikidata. Please take a moment to read the Wikidata:Administrators page and watchlist related pages (in particular Wikidata:Project chat, Wikidata:Requests for comment, and Wikidata:Administrators' noticeboard), before launching yourself into page deletions, page protections, account blockings, or modifications of protected pages.

Please feel free to join us on IRC: #wikidata-admin @ irc.freenode.net. If you need access, you can flag someone down at #wikidata @ irc.freenode.net. You may find Wikidata:Guide to Adminship to be useful reading. You may also want to consider adding yourself to meta:Template:Wikidata/Ambassadors, and to any similar page on your home wiki if one exists. (Check Wikipedia:Wikidata/Wikidatans (Q14964498).)

Please also add/update the languages you speak to your listing at Wikidata:List of administrators. You may also like to add your username to this list if you would not like that items you delete at RfD get marked as deleted automatically. Again, welcome to the admin corps!

--Lymantria (talk) 20:43, 7 February 2020 (UTC)

Adding wikidata items for sandbox articles

Hi @Mike Peel: Not sure whether I am in the right spot. I was hoping to speak with the @Pi bot:. It is surely not appropriate to be creating wikidata items for sandbox entries as Pi bot recently did for MargaretRDonald/sandbox/François Pellegrin (Q84532260). If Pi bot is your bot could you please adjust it so it does not create such silly, unnecessary and inappropriate items. I have deleted the contents of the sandbox and have deleted the wikidata link to the now non-existent article. If you know how to delete the wikidata item, I would very much appreciate its deletion. MargaretRDonald (talk) 20:24, 7 February 2020 (UTC)

@MargaretRDonald: You created your draft in article spare rather than user space - it was at en:MargaretRDonald/sandbox/François Pellegrin rather than en:User:MargaretRDonald/sandbox/François Pellegrin. As such, the bot couldn't tell that it wasn't an actual article. I've now deleted both the item here and the blanked page on enwp, if you use user space to create drafts in the future then the bot shouldn't do this again. Thanks. Mike Peel (talk) 09:23, 8 February 2020 (UTC)
@Mike Peel: Thanks, Mike. I thought I was in my user space. Thanks for the deletions. MargaretRDonald (talk) 17:29, 8 February 2020 (UTC)

Pi Bot

Q84986188

Hallo Mike,

why do you create a new items seconds after the creation of a new article? In many cases these articles may not survive due to several reasons. In this case the notabilty might be questioned due to his age. --Bahnmoeller (talk) 10:00, 13 February 2020 (UTC)

@Bahnmoeller: The script runs hourly, it's unusual that it catches an article within seconds. It checks to make sure that the article hasn't been created by a new editor (<200 edits), and isn't currently nominated for deletion. In this case, the item seems to be fine, both here and on dewp. That said, I want to investigate cases where this script hasn't worked so well (articles or items deleted) to improve it, so if you can point me towards problematic items/articles then I'd appreciate it. Thanks. Mike Peel (talk) 17:25, 14 February 2020 (UTC)

Occupation

Hi there, your bot got the wrong sport when it did this edit. I'm not sure how it thought she is an Association footballer, there is nothing in the article to suggest that. The-Pope (talk) 15:17, 17 February 2020 (UTC)

@The-Pope: It uses the Wikidata information about the categories to get the occupation. In this case Category:Australian rules footballers from Victoria (state) (Q8278102) had an error in it, which is now fixed. Thanks. Mike Peel (talk) 15:21, 17 February 2020 (UTC)
Great, thanks. I'd seen the error before, but didn't raise it, just fixed it. Glad you found the root cause. Cheers, The-Pope (talk) 15:26, 17 February 2020 (UTC)

Teatro Griego in Cordoba

One is in Spain, one is in Argentina., and you added the category to the wrong one. Of course, these mistakes can happen easily, but in this case I had specifically added a "different from" property to both items to prevent those kinds of errors. If the edit is done by some automatic help, maybe it would be advisible to put in a stop sign as soon as a "different from" appears in an item. Regards --Anvilaquarius (talk) 19:05, 22 February 2020 (UTC)

@Anvilaquarius: Thanks for catching that - and sorry for making the mistake! Mike Peel (talk) 19:09, 22 February 2020 (UTC)

Pi bot: lists of people deceased

Hi Mike,

Your Pi bot seems to be struggling with lists like Deaths in February 2020 (Q83274665) and Deaths in March 2020 (Q85544313). Would it be possible to make sure Pi bot does not state that this item is a human (Q5) and not adding a date of death (P570) to it?

This lists are made every month. Adding a follows (P155), point in time (P585) and part of (P361) would be fine, and labels in many languages could be made from similar lists in previous years (internationalisation of bots). I could work on that as well.

Thanks in advance, RonnieV (talk) 09:11, 20 February 2020 (UTC)

@RonnieV: It seems to be picking them up from the French Wikipedia, where the lists are in the same categories as the people that died. I'll try to add an exception for that into the code, so it won't create the items. Thanks. Mike Peel (talk) 09:23, 20 February 2020 (UTC)
Hello Mike, Thanks for looking at this too. It doesn't happen to often, just once or twice a month, but as my bot looks at all birth and deceases, these do show up while they should not be there. I'm sure you find a way. RonnieV (talk) 16:56, 20 February 2020 (UTC)
@RonnieV: I've added an exception if the frwp title includes "Décès", diff. Let's see if that fixes this - please let me know if you spot any other items that are created but shouldn't have been. Thanks. Mike Peel (talk) 17:21, 21 February 2020 (UTC)
Hi Mike Peel, Thanks. And no worries, I will let you know. RonnieV (talk) 09:39, 28 February 2020 (UTC)

SS Badger

Ok, but this ship have a heritage recongnition who was copied from Pi Bot (not that this ref num is also wrong), should we mo the heritage recognition categories on Commons to the c:Category:IMO 5033583? Since the ship will keep it except in case of demolition. --Fralambert (talk) 12:58, 24 February 2020 (UTC)

@Fralambert: Ah, sorry, the bot shouldn't have added the ID to the category item. I'm preparing a new version of the infobox to show more in the ship name categories, that's probably the best place for it, although I haven't yet decided whether it will show IDs. Demo is now in commons:Category:Badger (ship, 1953), but it needs some code tidying. Thanks. Mike Peel (talk) 15:07, 24 February 2020 (UTC)

Geburtsdaten

Könntest du endlich mal das genaue Geburtsdatum eintragen statt nur das Jahr? --Bahnmoeller (talk) 07:58, 5 March 2020 (UTC)

@Bahnmoeller: The bot adds full birthdays when it can, if you can point to examples where they are in the article (in a semi-structured way!) and the bot isn't finding them, then I can try to modify the code. Thanks. Mike Peel (talk) 08:15, 5 March 2020 (UTC)

Look at: Q86920665 de:Joachim Hösler

Joachim Hösler (* 22. September 1961) ist ein deutscher...

{{Personendaten |NAME=Hösler, Joachim |ALTERNATIVNAMEN= |KURZBESCHREIBUNG=deutscher Historiker und Hochschullehrer |GEBURTSDATUM=22. September 1961 |GEBURTSORT= |STERBEDATUM= |STERBEORT= }}

The "Personendaten" are suggested by a bot, a human user just has to confirm. So I think your bot can digest the first line of an article as well. --Bahnmoeller (talk) 08:41, 5 March 2020 (UTC)

@Bahnmoeller: Personendaten was added after the bot looked at the article, otherwise it would have used that. Is there a standard (and uncontaminated) pattern for birth/death days in the first sentence on dewp? Could you point me to the bot making the suggestions, so I can look at how it's doing it? Thanks. Mike Peel (talk) 08:50, 5 March 2020 (UTC)

NRHP

It looks like your bot is assuming that the presence of {{NRHP}} on a Commons category is an assertion that the category corresponds to the NRHP designation. It does not. By longstanding understanding on Commons, that template is also used on any subcategory that is composed mainly of content useful to illustrate that particular designated place. So, for example, a category for a building that is in a designated historic district might have the {{NRHP}} tag for that district.

This template is useful mainly because it links to the National Parks Asset record, things like https://npgallery.nps.gov/AssetDetail/NRIS/80004009.

I see that in (understandable) frustration with having to repeatedly undo the actions of your bot, people are removing the template on Commons. I really think a better solution is for your bot to stop making a wrong presumption about the meaning of the template. It can certainly prepare a list for human review, but currently it is drawing a wrong conclusion.

(Also -- different issue, probably not your affair, unless you have ideas -- some users are removing the template even from categories that really do correspond exactly to the NRHP designation once they see something visually equivalent show up in the Wikidata Infobox. Again, we lose the link to the National Parks Asset record. Any idea how we might address that? Any easy way to make that link available in the Wikidata Infobox?)

I don't maintain a watchlist on Wikidata, so please ping me if replying. Thanks in advance for any help. Jmabel (talk) 20:18, 7 March 2020 (UTC)

What I'm hoping we can end up doing is to add a parameter to commons:Template:NRHP to indicate if a category on which it is placed does not correspond to the whole listed entity; then the bot can see that in the template and know not to indicate in Wikidata that this is an NRHP ID for this category. - Jmabel (talk) 01:02, 8 March 2020 (UTC)

@Jmabel: I find that convention confusing, if the category is not about the NRHP listed place then why does it have the template? If it's part of a wider record, then I think it should be fine to have the NRHP number in the multiple Wikidata entries. In general, the template should no longer be needed in category space, as if it's correctly placed, then the number should be on Wikidata and in the infobox anyway. The link to the NPS website is in the infobox - looking at commons:Category:Dixie Highway, which currently has both the template and the infobox, the link looks the same. Am I missing something there? Thanks. Mike Peel (talk) 09:30, 8 March 2020 (UTC)
  • if the category is not about the NRHP listed place then why does it have the template? For the same reason that we put the template on individual photos: that they provide a good illustration for the topic. Remember that Commons templates aren't necessarily part of some deep ontology, just matters of convenience. Typically, the subcategories that have this are contributing properties of a historic district. Sometimes an NRHP designation will (often almost arbitrarily) pull together, say, three buildings of similar historical background that may not even be physically adjacent. There may not be any such thing as a photograph that would show all three buildings.
  • I think it should be fine to have the NRHP number in the multiple Wikidata entries. I guess that is for Wikidata to determine; I would think not, because Wikidata lists it as an identifier.
  • Am I missing something there? I think you are, but I think I also was. I see now that if I go down to the Authority Control section of the Wikidata box, then I get the same thing as from the template. Cool. I had not looked at that part. But I don't think we want that to be displayed that way except for the top-level category, which is why I'd like to add a parameter to the template for lower-level categories and have your bot skip them. - Jmabel (talk) 16:30, 8 March 2020 (UTC)
Hi, sorry for interrupting. First Jmabel was mentioning this change in made on Commons, mostly because Pi Bot will copy it to Wikidata. I deleted it because it was just a contributung property of Union Depot – Warehouse Historic District (Q49862016) who have alreay the refnum. One possible solution would be to modify Wikidata infobox so it could display the number of the historic district in the infobox. Is that possible? --Fralambert (talk) 22:48, 8 March 2020 (UTC)
@Fralambert: For the district itself it's there in the Authority Control section, but it's really hard to find if you don't know to go looking for it there. - Jmabel (talk) 05:27, 9 March 2020 (UTC)

Emergency medical services in France

"Emergency medical services" is the equivalent of "aide médicale urgente". --Bichatian (talk) 10:56, 11 March 2020 (UTC)

Thank you for participating in the FindingGLAMs Challenge!

Thank you for participating in the FindingGLAMs Challenge!
By improving information about GLAM institutions on Wikidata, you made the Wikimedia projects better for everyone!

Alicia Fagerving (WMSE) (talk) 14:08, 16 March 2020 (UTC)

Hi Mike, [7] is a bit unfortunate. Maybe you can do a run to adding missing Commons category (P373) to items that have topic's main category (P910) with Commons category (P373)? See for example https://w.wiki/KaL . Just ignore cases with constraints like Cheongju Han clan (Q12619259). Pretty sure that can also be done in SPARQL, but that will probably make the query time out. Multichill (talk) 21:29, 16 March 2020 (UTC)

@Multichill: I'd rather we just delete Commons category (P373) and use the sitelinks nowadays. I try to fix Commons category (P373) values when I can easily do so, and remove bad ones when I can't, but I don't actively add new values. I think @Jheald: and others add them via QuickStatements though. Thanks. Mike Peel (talk) 21:33, 16 March 2020 (UTC)
I haven't done anything on P373s for I'd guess about 18 months now. Jheald (talk) 21:37, 16 March 2020 (UTC)

Pi bot adding duplicate assertions

Here is a case where Pi bot added the P31=human and sex=male assertions twice. Bovlb (talk) 15:08, 19 March 2020 (UTC)

@Bovlb: Sorry about that. I suspect that the bot was somehow running twice at that time, perhaps because of the server slowdowns we've been having (it runs much slower when that happens, to avoid adding to server load). 5 March was a while back, though. If you spot any more examples, let me know, and I'll look into running the script less often. Thanks. Mike Peel (talk) 09:27, 20 March 2020 (UTC)

Dann mach auch den ganzen Rest. Vater, bruder, Vereinszugehörigkeit, Geburtsdatum (nicht Jahr), Geburtsort.... --Bahnmoeller (talk) 17:27, 28 March 2020 (UTC)

@Bahnmoeller: Sorry, I'm not sure if I translated that right. Is there a problem? It would be good to add the rest of the information you mention as well, but it's not easy as it's not structured in the article (it's not in the set of categories the article is in, persondata, etc.). Thanks. Mike Peel (talk) 17:34, 28 March 2020 (UTC)
If you unleash you Bot too early, nothing could be prepared in these 30 seconds. What your bot does is less then I am able to do with one click - but I prefer to add more and then I am running into a conflict with some stupid bots. --Bahnmoeller (talk) 17:41, 28 March 2020 (UTC)
It looks like the bot ran 20 minutes after the creation of the article? It must have been unfortunate timing if you ran into a conflict with it here. Thanks. Mike Peel (talk) 17:49, 28 March 2020 (UTC)

Please keep an eye on Pi Bot

hi, so I saw this very peculiar thing on https://commons.wikimedia.org/wiki/Category:Fiddles where the Wikidatabox was from "bluegrass" (Q213714) instead of fiddles (Q510487) it took quite a lot of research but it seems the reason was https://www.wikidata.org/w/index.php?title=Q510487&type=revision&diff=1145412371&oldid=1144849499 / https://www.wikidata.org/w/index.php?title=Q8303286&diff=prev&oldid=1145412430

I managed to fix this by removing (Q8303286) from fiddle (Q510487) and Category:Fiddles from Bluegrass fiddlers (Q8303286). I don't know why Pi bot did these changes, looking at https://www.wikidata.org/w/index.php?title=Q8303286&action=history it's possibly because someone added fiddle (Q510487) as main category topic to Bluegrass fiddlers (Q8303286), but some kind of double check should be in the bot when there are these "contradicting" statements. Thanks. CatQuest, The Endeavouring Cat! (talk) 11:25, 30 March 2020 (UTC)

@CatCat: Yes, that was due to bad data in Category:Bluegrass fiddlers (Q8303286), thanks for cleaning it up. It's likely that there will be more similar issues in the near future, as Pi bot adds the reciprocal topic's main category (P910) values, but I can't think of a way to avoid these in the first place - I think they need human cleanup when spotted. At least the bot edits make the bad cases more obvious so that they can be manually fixed. Thanks. Mike Peel (talk) 11:30, 30 March 2020 (UTC)

Year-only dates are imported wrong from the German Wikipedia

If the birth date is only given as a year, then it is imported wrong. In this particular case 1960 is imported as 196. --Zvpunry (talk) 20:47, 28 March 2020 (UTC)

@Zvpunry: That was imported because the article was in "Kategorie:Geboren 196". Fixed now. Thanks. Mike Peel (talk) 20:53, 28 March 2020 (UTC)
Ah, I missed the category and only looked at the structured "Personendaten" template. Sorry for the noise and thank you for pointing it out :) Zvpunry (talk) 21:40, 30 March 2020 (UTC)

Pi bot

Would it be possible to make it so your bot doesn't violate Abuse filter 55 as it did here? --Trade (talk) 22:05, 9 April 2020 (UTC)

@Trade: Huh, that's weird vandalism. I've deleted the item, I'll look into it more tomorrow. Probably it needs a check to make sure that dob nor dod are in the future... Thanks. Mike Peel (talk) 22:10, 9 April 2020 (UTC)

[8] The date was in the article en:Benedikt Kristjánsson, but only in a hidden comment; the infobox had been copied from another article. Peter James (talk) 15:09, 11 April 2020 (UTC)

@Peter James: I'm mostly on a wikibreak this week, but as a quick reply, the bot can't distinguish between comments and article text. The easiest way to avoid this is to not use hidden comments. Thanks. Mike Peel (talk) 20:25, 13 April 2020 (UTC)

E.g. this bot edit is wrong. When a category is based on the intersection of two or more topics, that category is usually not the main category for any of those topics individually. Please revise this bot. Thanks, Swpb (talk) 15:21, 13 April 2020 (UTC)

@Swpb: The bot follows category's main topic (P301) not category combines topics (P971) to add reciprocal values. I think @Infovarius: fixed the bad data that led to this edit [9]. Thanks. Mike Peel (talk) 20:23, 13 April 2020 (UTC)

Pi bot malfunction

Hello,

I have noticed that yesterday your bot was creating a handful of duplicate items (in that case both created by the bot) and adding duplicate (or even more) statements to items (creations or already existing) like in A L Eryomin (Q91134578), Augie Grill (Q91139095), Molly Huggins (Q91139834) etc. Could you please check what went wrong and maybe work on an efficient way to fix the duplicates? --Kostas20142 (talk) 15:30, 20 April 2020 (UTC)

@Kostas20142: Thanks for letting me know. I think it's because the server load yesterday was high, and then suddenly dropped, so multiple copies of the bot were then running through the same items. I'll have a think about how to better schedule it to avoid that. I think there's a different bot that removes duplicate values (perhaps KrBot - @Ivan A. Krestinin:?), so they should resolve themselves. Was it just duplicate statements, or also duplicate item creations? Thanks. Mike Peel (talk) 15:38, 20 April 2020 (UTC)
There were some duplicate item creations too, however they seem to have been merged manually. Thanks for looking through this. --Kostas20142 (talk) 17:54, 20 April 2020 (UTC)

interwikiextra

Hi Mike! I see from the thread above that you have an interest in a programmatic approach to generating fuller lists of interwiki links on client projects. I was wondering if you had any thoughts on the issue I raised here. (I was a little disappointed that it generated no discussion, but perhaps it was too niche for a general audience, or maybe it was the lack of a specific proposal.) Cheers, Bovlb (talk) 15:41, 22 April 2020 (UTC)

@Bovlb: I don't think I have an answer for you. The way it works for Commons now is that the equivalent of 'interwiki extra' is auto-included through the wikidata infobox when it detects a category's main topic (P301) or list related to category (P1753) value - also said to be the same as (P460) I think but I haven't double-checked that recently. You could perhaps do something similar through a common template on other wikis, perhaps using something like said to be the same as (P460). The other option would be to create redirects on the different wikis to the other topic, and then sitelink the redirects here, but some editors don't like links to redirects here. Thanks. Mike Peel (talk) 16:37, 22 April 2020 (UTC)
Thanks. I hesitate to rush to solution ideas before defining the problem properly, but I see two possible proposed implementations:
  1. Create a new property, say "fallback item for interwiki links". If item A has that link to item B, then user reading a Wikipedia article corresponding to item A would see interwiki links from item B when there are no links for that project/language on item A. We could then write tools to find and propose places to add this link based on other properties, or on interwikiextra usage in projects.
  2. Designate certain properties (such as those you list above) as being used for this purpose.
Obviously the two approaches could also be combined. Cheers, Bovlb (talk) 16:46, 22 April 2020 (UTC)
@Bovlb: (2) will be a lot easier than (1), both to set up and to maintain in the long run. Thanks. Mike Peel (talk) 16:50, 22 April 2020 (UTC)

P971

This property is used for the intersection of two or more topics. It is not used for a single topic. --EncycloPetey (talk) 19:53, 22 April 2020 (UTC)

@EncycloPetey: This is related to Category:Prometheus Bound (Eschyle) (Q21197387), right? It originally had category's main topic (P301)=Prometheus Bound (Q846870), and when pi bot added the reciprocal topic's main category (P910) value you reverted it, so I changed it to category combines topics (P971)=Prometheus Bound (Q846870) on the basis that it looked like it combined a book item with a translator that I didn't know the QID of. You've now changed it to category's main topic (P301)=Promètheus enchaîné (Q23308171), which looks like it's probably right? And you added the return topic's main category (P910) value, so all is now good? Thanks. Mike Peel (talk) 20:05, 22 April 2020 (UTC)
Yes, I had corrected Pi bot's edit to use the data item for the De Lisle translation, which is what the category on Commons if for. It contains the pages of De Lisle's translation. The Category name is the likely source of the confusion, because all editions of Prometheus enchaîné are by Eschyle. This particular is different because it was translated into French by Charles Marie René Leconte de Lisle, and a user/editor will not see that unless they follow the link. --EncycloPetey (talk) 23:47, 22 April 2020 (UTC)

Unwanted moves by Pi bot

Hi, Pi bot just performed some moves that trouble me: I linked new Commons-categories to Wikipedia-lists, yet Pi removes the links and links the Commons category to a Wikipedia category (f.i. this and this, but several more like this) - please stop this immediately. Eissink (talk) 06:39, 22 April 2020 (UTC).

Over the last week, I have spent a lot of time creating new categories on Commons, all in line with the above mentioned Category. If you want to just remove my work by bot, I would be totally unhappy, but it's going on and on. Eissink (talk) 07:14, 22 April 2020 (UTC).
@Eissink: Yes, the standard is that the commons category links go on the category item where one is available, and from the list item you can find them through the category related to list (P1754) link, the same as topic items have category's main topic (P301) links. to category items See User:Mike Peel/Commons linking for background. For example, en:Template:Commons category will follow category related to list (P1754) to find the commons sitelink for a list, but for a category it doesn't expect that - so en:Category:Mayors of Arnhem is currently in a maintenance category as it can't find the Commons sitelink on Wikidata. Thanks. Mike Peel (talk) 07:15, 22 April 2020 (UTC)
Thank you for answering, but I'm not experienced enough to follow what you say. Besides, I just want the Commons-categories linked to the lists, which totally list the items in the category, I don't want to link to another (incomplete) category on Wikipedia. When you say that you follow 'the standard', do you also mean to say that my construction (from which I made hundreds in the last week, like these) is forbidden and that I have no right to stop you from destroying my work, and that you will keep your bot going? Please stop immediately, I keep getting messages of destruction. Eissink (talk) 07:27, 22 April 2020 (UTC).
@Eissink: Are the moves causing a practical problem somewhere that I can help fix? I know it's a slight inconvenience to have to go to the category item to see the commons links here, which I can't do much about, but on Commons or Wikipedia I might be able to do something if there's an improvement that can be made? (e.g., I'm the maintainer of the infobox on Commons). Thanks. Mike Peel (talk) 07:31, 22 April 2020 (UTC)
I think it is totally logical and practical to link a commons category to a fully compatible list of items, which is what I had done. I'd like to maintain that link, and you can do something about it: just stop overriding my edits. Frankly, I don't really see the problem, since the majority of Commons Categories are linked to a Wikipedia page and not to the corresponding Wikipedia Category, yet somehow you try to tell me that what I did and what I consciously constructed is not allowed. That is how I see it, and I still don't see the ground you claim to act on. How do you see this all? Eissink (talk) 07:39, 22 April 2020 (UTC).
@Eissink: The central issue is that you can only have one sitelink from Wikidata to a Commons category, yet we essentially have three types of items here that may want the same sitelink: 'list', 'topic' and 'category' items (e.g., 'List of Mayors of X', 'Mayors of X', and 'Category:Mayors of X'). Since it's a commons *category* the convention has ended up being that the commons category sitelink goes with the 'category' item here, and then you use category related to list (P1754)/list related to category (P1753) to link that to the list item, and category's main topic (P301)/topic's main category (P910) to link the 'topic' item (although not all of these are necessarily present, so sometimes you'll only see parts of this, and the commons category can go on a topic/list item where there isn't a category item). On the Commons side of things, commons:Template:Wikidata Infobox follows these links to display the information it does (perhaps too well - you might not even know that there's a category item that's in use...), and at least on enwp en:Template:Commons category does the reverse to find the sitelink. It's a bit complicated, but it seems logical to me, and it generally works. Pi bot helps maintain this system, and the edits you've reverted all seem to have been OK in this scheme. Thanks. Mike Peel (talk) 07:50, 22 April 2020 (UTC)
What you like to call an "ended up convention" doesn't seem to reflect reality – what about categories like c:Category:Psychology and c:Category:Philosophy: they are not linked to the matching Wikipedia categories but to the topics (as they should), and there's many examples like this. Eissink (talk) 08:02, 22 April 2020 (UTC).
@Eissink: c:Category:Psychology is sitelinked to Category:Psychology (Q1983760), which then has topic's main category (P910)=psychology (Q9418), and c:Category:Philosophy is sitelinked to Category:Philosophy (Q1983674), which then has topic's main category (P910)=philosophy (Q5891). Thanks. Mike Peel (talk) 08:06, 22 April 2020 (UTC)
Maybe we talk about different things, but what I mean is that this Category:Philosophy page has, in the left side bar, a link to this 'Filosofie' article (and many links to other language versions, of course). I don't see a link to the Wikipedia Category on that Commons Category page, no matter how things are connected on Wikidata.
What I wanted, and what I had created, was also a link in the left bar of the Commons Category 'Mayors of'-page to the Wikipedia-page with a corresponding article, be it a list in these cases. Eissink (talk) 08:13, 22 April 2020 (UTC).
As I said, "perhaps too well - you might not even know that there's a category item that's in use..." For topic items, the interwikis are added by c:Module:Interwiki, which the infobox calls, but I'd forgotten that the same didn't apply for lists. I've requested a change to the Lua code at c:Module talk:Interwiki, and for the infobox this change would auto-call it where needed. It may take a day or two for the changes to be applied (and you shouldn't keep on reverting the bot in the meantime!) Thanks. Mike Peel (talk) 08:19, 22 April 2020 (UTC)
If this means (please confirm) that my edits will (eventually) hold, then of course I am grateful. In other words: have we accomplished an improvement? Eissink (talk) 08:23, 22 April 2020 (UTC).
@Eissink: I've reverted to the bot edits for c:Category:Mayors of Arnhem, figured out the change needed in Module:Interwiki, and am using the sandbox version of the infobox. You should now see 4 interwikis: 2 to lists, and 2 to categories (since lists don't exist in those languages). Is that what you are after? Thanks. Mike Peel (talk) 08:37, 22 April 2020 (UTC)
That is – at first glance – even better than what I was after, since I sought to display the interwikis to lists only. So this seems an improvement, though it is a bit strange to see interwikilinks to different entities (lists ánd categories). It is certainly satisfying, so thanks! Eissink (talk) 08:44, 22 April 2020 (UTC).
@Eissink: OK, see how it goes, and let me know if any other changes are needed. Note that you'll need to use the sandbox version of the infobox in the commons categories for now ({{Wikidata Infobox/sandbox}}), I'll update the main version later today. The bot should automatically redo its edits at some point. Thanks. Mike Peel (talk) 08:59, 22 April 2020 (UTC)
Okay. Sorry for the trouble, thanks for resolving. Greeting, Jurgen Eissink (talk) 09:27, 22 April 2020 (UTC).
@Eissink: Just to close this off, the update to the infobox on Commons is now live, let me know if you spot any issues. Thanks. Mike Peel (talk) 16:49, 22 April 2020 (UTC)
Hello, Mike. I'm not sure if it is part of the plan, but Pi bot is starting to delete Wikidata-Commons connections again, in favor of the Category-Category interwiki's (f.i. here and here). I'm afraid things are starting over again, so that's why I inform and ask you. Please let me know to do, whether I should undo Pi bot again or just wait. Thanks, Eissink (talk) 18:08, 25 April 2020 (UTC).
@Eissink: Part of the plan, I see the list interwiki at c:Category:Mayors of Coevorden still. Thanks. Mike Peel (talk) 18:11, 25 April 2020 (UTC)
Ah, I see. Thank you, Eissink (talk) 18:18, 25 April 2020 (UTC).

Sorry to bother you again, but in the latest run of Pi bot the connections with the list seem to get lost. I'm getting a bit desperate, also because an other user yesterday started to undo my edits. Could you please tell me whether what I do is somehow forbidden, is there a reason for me to stop making those connections? I really don't get it and I put so much effort in it already. From an epistemological point of view I see no problem at all, and in terms of ontological relations my construction is certainly richer than the category-category connection alone, not to mention the practicality of having a direct link on Commons, which was the main initial reason. Eissink (talk) 18:15, 27 April 2020 (UTC).

@Eissink: Can you point me to some examples where you're still having problems? While making the connections is important (and please keep adding new connections), there is a standard way of doing them such that categories link to categories, and lists/articles link to the category items. Thanks. Mike Peel (talk) 18:21, 27 April 2020 (UTC)
But what's the use of adding connections when they get undone? And the recent struggle with User Cycn, who just reverted because he hates me and nobody is doing something about it, is really, really depressing and tiresome. In the last hour or so there have been many changes made, f.i. here c.q. here and here c.q. here, but there's many more already, as you can see in Pi bot's history. The links to the list are just gone. Eissink (talk) 18:31, 27 April 2020 (UTC).
@Eissink: Add the sitelinks to the category items, and connect the list items to the category items using list related to category (P1753)/category related to list (P1754), and there should be no problem. Moving the sitelinks from category items to list items isn't helpful, but adding new sitelinks is useful. Thanks. Mike Peel (talk) 18:36, 27 April 2020 (UTC)
I understand from what you say that I should stop doing what I have been doing, but I have not been moving sitelinks from category items to list items, because the sitelinks on the category items didn't exist, because I created new categories. You say there should be no problem, yet the links from Wikimedia Commons to the lists on Wikipedia are gone, at least the ones that Pi bot has disconnected. Eissink (talk) 18:48, 27 April 2020 (UTC).
@Eissink: OK, please keep adding the new links, but please stop objecting to them being moved to category items. "the links from Wikimedia Commons to the lists on Wikipedia are gone" - they should still be present, and I've confirmed that with the examples you gave above - the links in the left sidebar still point to the lists after the moves. Am I missing something in those examples, or if not, are there other examples I should look at? Thanks. Mike Peel (talk) 18:57, 27 April 2020 (UTC)
Woerden, Gennep, Oost Gelre, Eijsden-Margraten in my browser, also when cache is cleaned, have lost the connection to the corresponding lists (Woerden etc.). If you want to compare, there's hundreds like these that are still fine: Teylingen. Yesterday all seemed fine, but not today.
I make categories like these, many of them – who makes clean Categories like these? not many people – and it's a lot of work, and I think I have something to say about how to connect them to Wikipedia, but it's your bot that messes it up, no matter what your intentions are. Why can't you just leave them alone? Certainly the problem is not with my edits, it's Pi bot that is tyrannical, as long as things don't change. If you don't understand what I mean, then please just let Pi bot ignore my edits. Eissink (talk) 19:12, 27 April 2020 (UTC).
@Eissink: Thanks for the examples. In the case of c:Category:Mayors of Oost Gelre the category was showing a cached version - clicking 'edit' and saving the page without changes (a 'null edit') refreshed it, and it's now showing the link to the list. The other three don't seem to have {{Wikidata Infobox}} yet - add that, and that should fix the issue, otherwise pi bot will add it automatically in the next few days. I'm sorry that all of this doesn't work immediately, but if you have patience then everything will work out OK. Thanks. Mike Peel (talk) 19:19, 27 April 2020 (UTC)
OK, I trust you in this. I will let it go. Thanks. Eissink (talk) 19:23, 27 April 2020 (UTC).

Hi Mike

You seem to be someone who understands how Wikidata 'links' (in the broadest terms) to other parts of the Wiki domain, as seen here. I wonder if you would be willing to help me get my head around what is happening, particularly Wikipedia. I created a category on WP the other day and was notified that a new linked item had been created on WD for it. I asked some questions about it, but I'm not getting very understandable responses. I know that there is a large swathe of WD data that originated in WP, and much of it was copied across using bots. That's correct, isn't it? Stuff like geodata/lat/longs/coord templates? Yes?

If so, is the link permanent, i.e. when the source {{Coord}} template is updated, is the WD data automagically updated too? If not, why not? Fob.schools (talk) 10:23, 29 April 2020 (UTC)

@Fob.school: I'm afraid this is a complicated issue. In general, Wikidata contains metadata about the categories and articles of the various language Wikipedias, and that information is used for the links between the different language versions, and particularly in the infoboxes in Commons categories. Information has been copied over from enwp (English Wikipedia) into Wikidata, but after the import it isn't necessarily kept in sync - it can be updated here on Wikidata, but not on Wikipedia, and vice-versa. It is possible to use templates on Wikipedia to import things like coordinates from Wikidata (just use {{Coord}} there without parameters, and it'll use Wikidata), but they are controversial.
Looking through your edits on enwp, you're adding values of UN/LOCODE (P1937) to the infoboxes. In the ideal case, you'd just add those to the Wikidata items here, and they'd auto-update everywhere else, but we're not there yet. I'd recommend adding them here as well as to enwp for now - adding them here is more sustainable in the long term, but enwp has higher visibility so is better in the short term. I hope that helps. Thanks. Mike Peel (talk) 19:49, 29 April 2020 (UTC)

Odd claim addition

Hi there, I wanted to call your attention to the edits made by Pi bot shown in this diff, as I'm not sure if I should have restored an older revision. I apologize if it's a mistake on my part. Cheers – Aranya (talk) 23:13, 1 May 2020 (UTC)

@Aranya: Thanks for letting me know. I've added an exception in the code so that it doesn't add birth or death dates that are in the future, which may also resolve @Trade:'s question from a while back [10]. Thanks. Mike Peel (talk) 14:28, 3 May 2020 (UTC)

Descriptions

Hi Mike. I have a little problem with my bot... Can you help me? The question is here. Thank you so much. --Vanbasten 23 (talk) 20:29, 4 May 2020 (UTC)

@Vanbasten 23: Definitely no '-er' in any of those cases. I'd suggest 'cyclist from Hong Kong', but "Hong Kong cyclist" also seems fine. Thanks. Mike Peel (talk) 20:40, 4 May 2020 (UTC)

Hi. I removed "instance of" "Country" becuse is redundant: Spain is already a instance of sovereign state (Q3624078) which is an in subclass of "Country" and therefore is redundant. Do yo agree? ⟨ RoberPL Dígame ⟩ 09:13, 6 May 2020 (UTC)

@Roberpl: Please see commons:Template_talk:Wikidata_Infobox#Location,_continent,_hemisphere,_error. I'm not sure if it's redundant or not, but it's good to have a clear indication that an item is an instance of a country, without having to go through subclasses. Thanks. Mike Peel (talk) 09:16, 6 May 2020 (UTC)
I see. Thank you for the information. ⟨ RoberPL Dígame ⟩ 10:00, 6 May 2020 (UTC)

Do we have consensus somewhere that list items (i.e. instance of (P31) Wikimedia list article (Q13406463)) or other "Wikimedia"-specific items (other than instance of (P31) Wikimedia category (Q4167836)) are allowed to be primary items for Commons category interwiki links on Wikidata? I noticed that you did that with list of Grange Hall buildings (Q6572374) but I've never seen it before. --Closeapple (talk) 21:21, 9 May 2020 (UTC)

@Closeapple: There's generally consensus to have Commons categories on topic items if the category item doesn't exist. For list items, consensus is less clear, but people still seem to do it a lot, and I can't see the harm in it if there's no corresponding topic/category item. If you create a category item and link it using list related to category (P1753)/category related to list (P1754) then Pi bot will automatically move the Commons sitelink to the category item. Thanks. Mike Peel (talk) 09:53, 10 May 2020 (UTC)

Bonagrazia de Bologne

Bonjour, Bonagrazia de Bologne (fr) = Bonagratia de San Giovanni in Persiceto (en) qui correspond à l'élément Q4941090 sur Wikidata. L'élément Q93955608 sur Wikidata n'est pas nécessaire. Cordialement --MOSSOT (talk) 17:21, 11 May 2020 (UTC)

@MOSSOT: Merged to / fusionné à Bonagratia de San Giovanni in Persiceto (Q4941090). Thanks/Merci. Mike Peel (talk) 17:51, 11 May 2020 (UTC)

Delete Spam Page

Q91833255 Self-promotion and Not Notable.157.47.143.182 09:40, 14 May 2020 (UTC)

Hotkey for Wikidata Infobox in edit window

Hi, Mike Peel. I thought I'd let you know that I requested a hotkey for easy insertion of the Wikidata Infobox template on Commons and it has been implemented. To me it is an improvement and I think it will be useful for others as well. The hotkey only shows up in Category pages, I believe, which is of course appropriate. Greetings, Eissink (talk) 09:31, 16 May 2020 (UTC).

Matching existing wikidata objects with unconnected articles

Hello Mike,

in the last days wikidata objects for humans have been created, which currently only contain the name and the GND, see:

  • Will it be possible for Pi bot to match articles, which will be created in the future to these existing objects? Or does this now have to be done manually (as for example with all articles, which have been created in the user namespace and have been moved to the article namespace lateron or added information like categories, infoboxes, templates, ... at a later point in time, ...).
  • Which are the criteria for Pi bot to decide, if an object and an article describe the same person (name + date/year of birth + optional date/year of death?).
  • Is it possible to match objects and articles based on various IDs, like VIAF, GND, IMDb, LCCN, Transfermarkt, Weltfussball, ... which are included in the articles (and the objects)?

Thanks a lot! --M2k~dewiki (talk) 10:01, 24 May 2020 (UTC)

@M2k~dewiki: The script that creates new items for humans:
  1. Runs a search for the person's name
  2. skips cases where more than 5 matches are found
  3. excludes matches that already have a sitelink
  4. skips cases where it finds a potential match without a birth date
  5. excludes matches with a birth date is more than 3 years different
  6. adds the sitelink if the birth year matches
  7. only if there are no remaining matches, then it creates a new item
In this case, I think the script will stop at #4, since the newly created items don't have birth dates in them - those cases will have to be manually resolved.
I do have other scripts that add sitelinks based on identifiers, but they were only written for importing commons category sitelinks based on IDs on Commons. I could try repurposing these to look through identifiers in Wikipedias, but there would have to be a straightforward way to find the cases where the article doesn't have a sitelink and has an identifier (e.g., a tracking category that contains articles with identifiers that don't have Wikidata items, or have identifiers that don't match Wikidata). Thanks. Mike Peel (talk) 10:16, 24 May 2020 (UTC)
GND is CC0 and has a linked data service from which the date of birth can be obtained. MrProperLawAndOrder (talk) 14:19, 24 May 2020 (UTC)
@MrProperLawAndOrder: Could you import them at the same time as creating the items? QS should support dates, you just need to be able to get the data into the right format to put into QS. Thanks. Mike Peel (talk) 14:21, 24 May 2020 (UTC)
Not so easily. I imported based on deutsche-biographie.de but would prefer to take the time information directly from GND. I am in contact with User:Epìdosis and he with User:Bargioni, the latter seems to have expertise with perl scripting and reading data from website, so that seems to be one good option to obtain the information. Anyway, the newly created items all have GND and can thus soon have a VIAF. Articles created in dewiki shall get a GND locally anyway and then it is easy to connect/merge them based on the GND. I don't know how often Pi bot works on articles that already have a GND. Last but not least, matching by GND - if the GND is correct - is much safer than by name+birthdate. An article creator can probably best tell which GND is the correct one if faced with multiple items having same name+birthdate. GND DB stores information about 12 mio+ humans, WD has not even 1 mio GND humans. Very soon, your tool may work much better than before, because the number of GND humans with rich and verified data will be increased a lot. Apart from User:Epìdosis, User:Kolja21 is involved in the DtBio centered GND human data upgrade. MrProperLawAndOrder (talk) 22:42, 24 May 2020 (UTC)
BTW: It would be a great help if a bot can connect the articles published in the Allgemeine Deutsche Biographie (Q590208) (ADB) with the person described by the article. Example:
The connection is shown by Deutsche Biographie (GND) ID (P7902). --Kolja21 (talk) 23:17, 24 May 2020 (UTC)
@Kolja21: I can see Deutsche Biographie (GND) ID (P7902) in Thomas Bomel (Q1320327), but I don't see the same ID in Bomel, Thomas (Q27562732), so I'm not sure how to find that entry based on the Deutsche Biographie (GND) ID (P7902) value? Unless you were meaning matching based on names and the existence of Deutsche Biographie (GND) ID (P7902) in the biography article? Thanks. Mike Peel (talk) 08:14, 26 May 2020 (UTC)
@Mike Peel: The GND is stored in the template:ADBDaten. Wikisource s:de:ADB:Bomel,_Thomas (template:ADBDaten), GND 135554322 = Thomas Bomel (Q1320327), GND 135554322. --Kolja21 (talk) 09:12, 26 May 2020 (UTC)

Back to what M2k~dewiki wrote and #4 in the reply. The new items are only additional items, not even with dates of birth and deaths yet. If before a match could have been made why is that prevented from the existence of new items in WD? There are millions of humans not yet in WD. Why is adding items stopping assignment? #2 "skips cases where more than 5 matches are found" why that? What does the amount of name matches to do with matching? Millions aren't in WD yet. MrProperLawAndOrder (talk) 22:59, 25 May 2020 (UTC)

Hello @MrProperLawAndOrder: I guess the bot tries to avoid creating duplicates for already existing objects (which might have to be merged manually afterwards). If the algorithm can not decided which object is the right one, it skips creating a new object and a human should decide, which object is the right one or if a new object should be created.

For example:

A new article de:Wolfgang Jahn (Fußballspieler) has been created. The algorithm would have to find out, which of these objects is the right one (or if a new object has to be created).

Q1658584 and Q66030227 already had sitelinks for the german wikipedia (it could also be a duplicate in the german language wikipedia, i.e. two articles for the same entity, which sometimes also is the case), while Q94938019, Q95243521, Q95341215 did not have any date of birth. The algorithm would have to decided if one of these objects should be connected to de:Wolfgang Jahn (Fußballspieler) or if a new object is necessary.

In my opinion, one of the most difficult tasks is to decide, if a new object is necessary or if it already exists, especially with different languages and alphabets (cyrillic, japanese, chinese, ...) for about 350-500 articles, categories, disambiguations, lists, templates, ... created every day in the german language wikipedia for example (also see Wikidata:Metrics), all of them needed to be connected to an (maybe already existing) object, while trying to avoid duplicates (or one or more "clusters" of objects in different languages, which actually describe the same entity, so the users cannot switch between all existing languages for the same article, but only a subset of them), also for different languages. --M2k~dewiki (talk)

M2k~dewiki, but for items from German Wikipedia duplicates wouldn't be a problem if they are created due to the new DtBio items having no time information. Since German WP adds GNDs the duplicates will be found via Property talk:P227/Duplicates#human and be merged. Also, why does the process skip if several names match, but not look at dates? This could be changed. The ~764000 DtBio humans are a subset of the 12000000+ GND humans, aggregated from dozens maybe 100+ research organisation in Germany, Austria, Switzerland. Re "350-500 articles, categories, disambiguations, lists, templates ... created everyday" - the DtBio items only affect articles, and only articles about humans.
I thought it would benefit the Wikimedia world to have all the DtBio humans in WD and to make sure they are of high quality. Phase one was to create the items. Now this is interrupted because you claimed that tools are broken and Jura1 jumped in based on this claiming useless etc. and is threating me. How many human items are created in dewiki per day? What fraction of these is affected by the DtBio humans? WD also helps to find errors in GND DB, most notably duplicates. Having the DtBio items in WD will make accelerate the data improvement for these GND humans and have a positive effect on knowledge management world wide. MrProperLawAndOrder (talk) 00:14, 26 May 2020 (UTC)

For the record: The thread was started 24 May. The opener created a low data item 25 May [11], having less data than the ones he complained about and no indication of planned enrichment - while for the ones he complained about such plans have been presented. Notably containing no year of birth, while the article and GND contain such a value. MrProperLawAndOrder (talk) 07:00, 26 May 2020 (UTC)

@MrProperLawAndOrder: The aim of bots in general is to reduce the amount of human work that needs to be done, so in this case edge cases are skipped to minimise creating duplicate entries that have to be human-merged later. I can tweak the code to skip more/less edge cases as needed, but the current settings mostly work OK as things stand. I don't have an issue with what you're doing, but modifying your approach so that you add more info to an item shortly after creating it would seem helpful. Thanks. Mike Peel (talk) 08:08, 26 May 2020 (UTC)
It was planned to first create all the relevant items from DtBio and then to enrich them. Now the task as been disrupted by M2k and Jura1 pointing to your tool. But M2k is "disrupting" your tool in the same way. And neither he nor you explained what actual disruption the new items cause. If a new Fritz Müller makes your tool stop, then one can also be happy that this was done, because the tool may have otherwise connected the wrong Fritz Müller, not "knowing" there could be another. The new items make edits by the tool safer. Could you agree with that statement? There is no data by M2k or you about how many cases are handled differently due to the new high importance items and especially not about how often things work worse than before. I get around 10 notifications for the new items every day when they are linked from other items. So, there is already the clear plus side of having them. @Kolja21, Epìdosis: your opinion? MrProperLawAndOrder (talk) 08:20, 27 May 2020 (UTC)
@MrProperLawAndOrder: It's not so much a problem for pi bot - it just edits less - as it is for humans that then have more items to look through to manually match or create new items for. My suggestion is that you try to work in batches - import a batch from DtBio, then expand those with information from elsewhere, before moving on to the next batch. I don't keep track of the number of cases that pi bot skips, but I just did a test run of the version that runs every hour, reporting skipped items, and here's what it came out with:
It turns out that I was wrong above - step #4 checks to see if the *article* has no birthdate (like Derry Brabbs here), but that is then followed by a check of whether the Wikidata item has a weird birthdate - which includes no birthdate in the item. Most of the cases were on enwp, but the last was from dewp, and is one of the items you created. Thanks. Mike Peel (talk) 17:03, 27 May 2020 (UTC)
If it edits less, maybe this is good, as it reduces chances for wrong matches, see Fritz Müller above. DtBio is already a subset of GND, to create further subsets and do mass operations (e.g. enrichment) on these is complicating the processes. Re Q94875330: articles in dewiki receive GND IDs early in article creation process, and the GND id was on the WD item. So, so far one case found where the bot skips but hundreds of users already benefit from the newly created items and soon they will have much more information. MrProperLawAndOrder (talk) 17:20, 27 May 2020 (UTC)
@MrProperLawAndOrder: I suggest that it's worth "complicating the processes" - at the very least, if you do that now, then you have practice in how to batch-create Wikidata items that are newly added to DtBio in the future. It's rarely possible to work on a complete dataset here, and it's worth designing your import process to bear that in mind. But this isn't my dispute and I don't want to get too far into it, I'm just providing input and suggestions here. Thanks. Mike Peel (talk) 17:25, 27 May 2020 (UTC)
I see no basis for your suggestion to complicate the processes, another user doing the actual work also preferred to finish one task before starting the next. Task 1 is to create new items to have the set ready that we plan to work on later. One cannot add parents to an item if the parents don't have items yet. Re "then you have practice in how to batch-create Wikidata items that are newly added to DtBio in the future." - the same is achieved already now. Task 0 was to add P7902 to existing items, then adding the missing items. MrProperLawAndOrder (talk) 17:37, 27 May 2020 (UTC)
@MrProperLawAndOrder: I'm only offering suggestions based on my experience here - it's up to you if you want to listen to them. All I can do beyond that is wish you well with editing here, and to say thank you for your contributions. Thanks. Mike Peel (talk) 17:39, 27 May 2020 (UTC)
I am listening. I did stop processes. Many items of structural need are missing, parents, spouses, children. But you provided no evidence that it makes sense. You didn't even answer regarding the potential benefits of avoiding wrong assignments. If you have any substantiated concerns please share them at the central discussion Property talk:P227#Create GND humans from Deutsche Biographie. Your bot just added several details to Q94875330 :-) Thank you!!! MrProperLawAndOrder (talk) 17:50, 27 May 2020 (UTC)
@MrProperLawAndOrder: Just to clear up two points: "You didn't even answer regarding the potential benefits of avoiding wrong assignments" - I provided info on how the bot already avoids wrong assignments before creating new items, and how your items would make it harder to find the correct item; would stop the bot from auto-creating new items in those cases; and that it can't match the article with your new item if that was the correct item to match it to. With Friedrich August Fritzsche (Q94875330), pi bot added more info after @M2k~dewiki: manually added the sitelink. Thanks. Mike Peel (talk) 17:55, 27 May 2020 (UTC)
You wrote "has weird birthdate" (in the article, the WD had no birthdate at all) as the reason for the article not being connect. So the existence of the item changed what in the bot process. Regarding the other points: no, you didn't show how the items disturbed the bot process and no, you didn't admit the new items could help to avoid wrong assignments. The bot does operate on names and dates and maybe other data, but not on IDs, so wrong assignments can happen all the time. MrProperLawAndOrder (talk) 01:48, 28 May 2020 (UTC)

Quote from Wikidata talk:WikiProject Authority control: "your Pi bot creates human items based on Wikipedia articles - if doing it from dewiki, could it also import VIAF and GND? MrProperLawAndOrder (talk) 15:28, 9 May 2020 (UTC)

Possible, but that bit of code is already quite complicated, so I'd prefer not to complicate it further. It sounds like there are other routes to import those values. Thanks. Mike Peel (talk) 15:57, 9 May 2020 (UTC)"

GND is a high value indicator to match articles about human from dewiki. Your bot not using it could be a reason for any mismatch from dewiki, since all items I create have it. If your bot could add the GND then via Property talk:P227/Duplicates users will take care of merges. MrProperLawAndOrder (talk) 01:42, 28 May 2020 (UTC)

Strange date of decease

Hello Mike Peel,

Can you trace why your bot made these two edits, based on this version of the English article about Neil Duffy? Neither this version (the first) nor any of the later say that Duffy died in 2020. All clearly state 2013.

Moreover, I think it would be good to have a bot not introducing values that apparently can not be right (or at least: known) at that time. Only someone who is committed on killing someone on a specific date in the future could know the planned date of decease, and then has to hope the planned victim does not die before.

Thanks for your time and improvements, RonnieV (talk) 15:49, 1 June 2020 (UTC)

@RonnieV: Pi bot did connect to the internet using a "skynet" wifi network in the past, perhaps that could explain it? ;-) More seriously, the problem seems to be the unusual format of 17 June {{death date and age|2013|1937}} - when I run a test on the current version of the article it still returns "2020-6-17". I thought I'd modified the code to not add dates in the future, [12], but that only checks that the year isn't in the future at the moment. I need to investigate and improve the code, I will do that soon. Thanks. Mike Peel (talk) 16:04, 1 June 2020 (UTC)
Thanks for the explanation, Mike. Future dates are relatively easy for me to see (RonnieBot does read and write deceases to the Dutch Wikipedia). It is harder for me to see if Pi bot (or others) do make similar mistakes on past dates (17 May 2020 would have been less visible). I hope you can trace the error and solve it.
Having a look at the templates documentation, it says that the first four parameters are required. Would it be a save option to ignore templates which do not have (at least) four parameters? Greetings, RonnieV (talk) 16:47, 1 June 2020 (UTC)
@RonnieV: I think I've fixed it with this edit. The code was returning a year of "e", so I now make sure the year is numeric before using it. In this case, that causes the code to fall back to the death date category, hence it would now make this edit. That's not ideal as it misses the day and month, but is better than before, and it would make the code more complex to add further checks for this edge case. Ironically, this would all be a lot easier if people put info into structured data *first*... One day, I hope. Thanks for spotting and flagging this error! Thanks. Mike Peel (talk) 17:48, 1 June 2020 (UTC)