[04:34:51] Danny_B, you were the biggest advocate of the commons datasets... and i haven't heard from you in ages! are you ok? [10:37:47] nikki: I had to dedeprecate so many statements and add end time [10:37:47] nikki: we've adjusted the tooltip hoping that would help a month ago or two [10:37:47] still seeing a lot of issues? [10:37:47] Lydia_WMDE: It's all paintings stolen around WW2 and returned to the Netherlands. https://www.wikidata.org/wiki/User:RKDdata is helping out with the weird things we find :-D [10:37:59] heh [10:38:38] would be good to know if it still happening now on a larger scale or if things got better [10:39:10] For me it was just a couple of users who did it to a bunch of paintings, easy to correct [10:39:19] Much harder when people remove the whole statement [10:40:22] I haven't checked lately (I've been avoiding it, I only get annoyed by it :P), but I can't imagine that a tooltip would change much :/ (I didn't even realise it had tooltips) [10:40:23] Infobox wants to show all information that is valid now, Wikidata item wants to show all data that was valid at some point in time. Those two concepts keep biting eachother [10:43:08] wtf https://www.wikidata.org/wiki/Q7769 [10:43:43] a bot changed the p131 statement to deprecated and added a different p131 statement which is somewhere inside the original p131 statement [10:43:54] that's not even old data [10:45:46] nikki: Can I use your optimization skills on http://tinyurl.com/jss5r6m ? :-) Started timing out :-( [10:46:37] I have 5 collections that use the same inventory number (every 10 years the collection was transfered to some new organization). This query is to hunt down duplicates [10:48:45] I've been having problems with things timing out too :/ [10:48:53] especially when I use "values", even if it's just two things >_< [10:52:22] yeah, getting rid of values makes it return in milliseconds for me - http://tinyurl.com/j9xlmgg [10:52:58] might need a distinct in there [10:53:28] Yeah, let's file a bug for this shall we? This is something the optimizer should handle [10:54:06] by "we" do you mean you or me? :P [10:54:36] I'll do it. Do you remember how to pull the optimizer info from the SPARQL engine? [10:56:09] https://www.wikidata.org/wiki/Wikidata:SPARQL_query_service/query_optimization says to use https://query.wikidata.org/bigdata/namespace/wdq/sparql?explain&query= [10:59:00] nikki: https://phabricator.wikimedia.org/T154280 [10:59:21] thanks [11:01:26] Lydia_WMDE: That's a fun project for Computer Science students for their BS/MsC project, improving the SPARQL optimizer. [11:02:26] yeah indeed. i'll poke Markus a bit about it. he might have students interested in that [11:06:30] multichill: if this id was removed from the other db, doesn't that mean it was once correct? https://www.wikidata.org/w/index.php?title=Q13816220&diff=377056611&oldid=376091171 [11:07:41] I also have no idea what to do with items like https://www.wikidata.org/wiki/Q18928548 where all the statements are marked as wrong [11:07:48] Haha, wondering that too. The only reason I didn't remove the links is that people kept adding them [11:09:04] RKD purged a lot of records that were not up to their standards. I marked all of these as deprecated to prevent mix'n'matchers from adding them [11:09:54] * nikki nods [11:10:16] Wouldn't be very nice of us to over our users broken identifier links [11:10:18] that's another reason I'd like an outdated rank, so old (but not incorrect) identifiers that got deleted or redirected could be marked as less-than-normal rank without using deprecated [11:10:50] The current description of the deprecated rank includes outdated information [11:12:18] yeah, I think that's part of the problem, because I don't think it was intended to include what most people consider "outdated" [11:12:39] And of course identifiers are a special class of properties [11:12:49] Hi nikki, I'm new here. Where do I see that Q18928548 statements are marked as "wrong"? [11:13:20] I tried to have an item deleted the other day because there was just nonsense in it, and since it was "used", someone restored all of the nonsense. [11:13:31] I'l love to at least mark it as "wrong" [11:14:04] (there is a statement that Wikimedia is a "category: day in carneval", absolute nonsense. [11:15:09] WiseWoman: do you have a link to it? [11:15:17] Just a moment, hi Lydia! [11:15:21] like "outdated knowledge", as I understand it, was only intended to mean "things we used to think we true but now realise never were" not everything which could be considered out of date [11:15:22] hey :) [11:16:21] WiseWoman: there's a little icon to the left of the values (e.g. to the left of "disease") and when the lower icon is grey, that means it's marked as deprecated [11:18:00] thanks, nikki! [11:18:12] WiseWoman / nikki : Check out the creator on https://www.wikidata.org/wiki/Q28062107 . We thought Johannes Vermeer, but turned out to be a forgery [11:18:31] I see the icons on https://www.wikidata.org/wiki/Help:Ranking haven't been updated :P [11:19:54] https://www.wikidata.org/w/index.php?title=Q28060447&type=revision&diff=421908319&oldid=421345710 [11:20:25] Lydia_WMDE: Oh, I see now, you have all of the Wikimedia categories as Items. It was https://www.wikidata.org/w/index.php?title=Q9135968&oldid=377842473 (someone put "Wikimedia" in as the label (which was wrong) now it's been fixed to have the label gone https://www.wikidata.org/w/index.php?title=Q9135968&oldid=421459161 but in the RfD Hakan-IST just said: not done, labels and description restored [11:21:14] cool [11:21:47] I play the Mix'n'Match game when I'm bored. Does anyone know if multiple people have to agree before something is added? [11:22:15] Just curious. [11:23:24] multichill: Thanks, that is an *excellent* example of something in RL (tm) that must be modeled somehow. [11:24:24] nikki, I always thought they were for moving the items in the order in which they are displayed :) [11:24:34] No, mix'n'match adds it right away so be careful with what you approve. It's broken at the moment by the way [11:24:37] And I wondered why they didn't work as I expected them to :D [11:25:09] multichill, it was broken last night, it was giving me the same nonsense to add to a gazillion items [11:25:15] yeah, that's what the icon suggests to me as well [11:26:40] I added some stuff to the Kingdom of the Netherlands and then didn't like the order in which it appeared (I had entered end date then start date for the monarchs, and that looked sick and I wanted that to then be start data then end date). Any way to change that? Not that it matters for a query. [11:28:27] multichill, shouldn't you have to add a time frame to https://www.wikidata.org/wiki/Q28062107 saying "Up until we thought this was Johannes Vermeer, from we are sure it is a forgery" and then include a link to some paper explaining it was a forgern? [11:33:41] multichill, is there a reason you don't have the name of the painting as a label? [11:36:42] WiseWoman: The source data is only in Dutch, so I added a label in Dutch [11:37:02] This is part of a batch of 1250 paintings I uploaded the day before yesterday [11:37:44] Very cool. I would have thought that if any label was given, it would be taken, but apparently only the English name is a label. Are there official translations of the names into English? [11:39:06] The RKD usually has records with English names, see https://rkd.nl/en/explore/images#filters[kunstenaar]=Meegeren%2C+Han+van [11:39:29] multichill, would it make sense to have something like "things depicted in the painting"? I can imagine someone searching for "some old painting with a cembalo" and being thrilled to find this :) [11:39:35] being able to query for when statements were last updated would be useful right now... trying to find recently changed deprecated rank statements... [11:39:38] we do have that [11:39:41] "depicts" I think [11:40:06] WiseWoman: Yeah, we do that, see for example https://query.wikidata.org/#%23defaultView%3AImageGrid%0ASELECT%20%3Fitem%20%3Fimage%20WHERE%20%7B%0A%20%20%3Fitem%20wdt%3AP180%20wd%3AQ1545193%20.%0A%20%20%3Fitem%20wdt%3AP18%20%3Fimage%20%0A%7D%20LIMIT%2051%0A%20%20%20%20%20%20%20%20 [11:41:53] http://tinyurl.com/j9jcqf3 in case that didn't work WiseWoman [11:42:32] That's paintings depicting the main church of Haarlem (where I'm from) [11:43:18] multichill, that's awesome - now we need a user interface for people who don't want to use SPARQL :) [11:45:50] SPARQL has quite a steep learning curve, but quite a few people already got the hang of it. You should give it a shot [11:46:42] See for example the things Jane made at https://www.wikidata.org/wiki/User_talk:Jane023#Paintings_by_decade [11:48:00] multichill, oh, I teach Semantic Modeling and we do SPARQL :) But I'm thinking of the normal people who want to find paintings of harpsichords. They can't learn SPARQL for just a simple search [11:48:58] That might make in interesting thesis for students (but I'm on sabbatical next semester, so this would have to wait). [11:50:03] Lydia_WMDE: I can't find any way to find recently deprecated statements or to quantify how many statements are being marked as deprecated so I have no idea if anything's changed :( [11:52:41] nikki, shouldn't that be possible running over the history? For all items from until if history deprecation added, print out? [11:53:53] nikki, yup, depicts does the trick. The painting now depicts a harpsichord, but I'm not too sure I can just translate the title from the Dutch ("Interior with a couple at the harpsichord") [11:54:18] it's theoretically possible, but the data isn't available in any easy to query way [12:01:14] nikki: yeah :( needs going through the dump i fear [12:01:39] * Lydia_WMDE hopes we can have a data anaylist for this kind of stuff at some point [12:08:13] nikki, you will want to be going through the history looking for stuff like this: https://www.wikidata.org/w/index.php?title=Q28062107&type=revision&diff=421629204&oldid=421629139 There's a tool called "Wikiblame" that you can use to pinpoint the exact point in time that a particular statement was added to a Wikipedia. This could serve as a basis for such a "deprecation-finder". [12:11:26] yeah, but I have plenty of things to do already :/ [12:11:38] Lydia_WMDE: Since many companies are screaming for data analysts and we can't produce them fast enough, you could maybe get the foundation to let you hire one trained analyst with the promise that their job is to train new data analysts by offering internships. I find Wikidata to be a really great took for getting people to understand what you can do with semantic data. [12:11:46] nikki Me too :) [12:12:42] Lydia_WMDE: That way the foundation could be contributing to the world by training more data analysts AND fixing up Wikidata at the same time. Although I do understand that it was hard enough to get the financing for Wikidata sorted out as it is :) [12:19:21] and I'm not really that interested in finding out whether it's changed, because it doesn't change anything for me... [12:19:58] you still can't edit a statement to say it's not preferred, you can only edit *other* statements to say that they are... but just because I know one statement is out of date doesn't mean I know which of the other ones aren't [12:23:27] nikki: Do what I do, set up a Listeria list to track it [12:25:29] I can't compare before the change and after the change though [12:25:49] multichill, Listeria? Do you have a link? [12:26:32] nikki, yeah, I've been experimenting with Wikiblame and searching in the revision history. Although the comparison shows "Deprecated rank", the way the data is stored is not searchable in this manner. [12:30:02] multichill, found it http://magnusmanske.de/wordpress/?p=348 There are far too many pages about Listeria the bacterium :) [12:49:13] Aleksey_WMDE: https://www.mediawiki.org/wiki/Wikibase/Programmer%27s_guide_to_Wikibase#Browser_Testing_for_Wikidata [13:48:54] I have an .ods with 3468 Spanish beaches to be imported to Wikidata [13:49:05] However, around a half of them already have an item [13:49:21] Which tool should I use? [14:08:39] abian: You need to cross reference them before importing. Did you already do that? [14:13:52] I have references for some properties which the Qs should have so that we can find them easily, but not the existing Qs themselves [14:15:47] I mean, I have how the beach is named (label) and a series of values for properties, some of which are Qs [14:17:57] But I've not determined if every beach exists or not in Wikidata, I only have an .ods by one side and a query by the other... [14:18:00] http://tinyurl.com/jertjo3 [14:20:41] https://nube.wikimedia.es/public.php?service=files&t=a0c23d50b6de668ef23dc69356994338 ← spreadsheet [15:46:53] nikki: https://www.wikidata.org/wiki/Wikidata:SPARQL_query_service/queries/examples#Painters_related_to_anonymous_works <- another fun optimization problem [15:47:06] If you remove the optional, it times out [16:35:06] How do I add a statement to an item? Can't find it on https://www.wikidata.org/wiki/Q2941 [16:47:44] Papierkorb: there should be an "add" link at the bottom of the statements, if you don't see it, try refreshing the page (sometimes it doesn't load properly) [16:51:34] nikki: Ah. Looks like JS went nuts, now everything works [19:38:29] wut, why is https://www.wikidata.org/wiki/Q16503 protected [19:39:14] Nemo_bis: 18. Nov. 2013, 21:23:39 Stryn (Diskussion | Beiträge) schützte die Seite Wikipedia:Projektdiskussionen (Q16503) ‎‎[edit=sysop] (indefinite) (To disable the automated page-move-change-sitelink that happens twice in a month) (Versionen) [19:39:35] doesn't explain anything [20:18:28] Nemo_bis: my understanding is that some projects move the page to archive it, causing edits like https://www.wikidata.org/w/index.php?title=Q16503&diff=13983404&oldid=13829324 [20:19:00] and protecting it prevents those edits, so the link stays correct [20:29:39] It would be easier to remove the links to those wikis and let the others live [22:43:50] https://www.wikidata.org/wiki/Q21286937 How do items like these happen, and why are they not deleted? [22:44:59] because they have sitelinks [22:45:05] we track them [22:45:23] once the problem is resolved on Wikipedia, we can resolve it on Wikidata too [22:55:33] that's not right [22:55:55] they won't be resolved on wikipedia because it's deliberate [22:56:21] and it happens because wikidata only allows one sitelink per wiki which doesn't reflect how a bunch of wikis work [22:58:29] nikki, right, and there is often multiple articles of some facet of life in some wikis that only have one in another. I've given up trying to add Wikipedia articles, as they are usually linked to some other item [22:58:55] * yurik pokes SMalyshev yet again :) [22:59:41] can someone with a good internet connexion sync https://tools.wmflabs.org/mix-n-match/?mode=sync&catalog=208 ? [23:00:23] the page is too heavy, doesn't load at mine [23:00:39] you shouldn't need to be oauth for the sync to work [23:00:58] Harmonia_Amanda: (Hi!), sorry, I've got a flakey mobile Internet connection at the moment [23:01:42] if you aren't oauth it's just appears as "auxiliary data matcher" [23:01:57] WiseWoman: thank you for answering me [23:01:59] there are issues with articles in different languages covering different aspects of things and working out how to link them on wikidata, but that item isn't that, it really is about two articles for exactly the same concept (written in different dialects or scripts) [23:03:19] nikki, I'm curious, as I'm from Berlin :) [23:04:32] nikki, https://frr.wikipedia.org/wiki/Berlin/mo is in a German dialect (Northern Friesian) [23:06:34] There are 9 subdialects given there, that is the /mo on the end of the article [23:06:43] yes [23:06:58] only one other (the sölring version) has an article, that's the one linked to the main berlin item [23:06:59] There is only one other subdialect given [23:07:01] :) [23:07:05] Ah, I see [23:07:20] But it it important to have the Mooring one linked as well? [23:07:37] There is a nice box on the page at Wikipedia that links to the other subdialects [23:08:39] the mooring version is missing all the interwiki links, it would also be more work to use wikidata on the mooring version because it's not directly linked to the proper berlin item [23:08:48] I see what happens, Wikidata puts in an extra item "permanent duplicate item" and then links it there, but if the other 7 subdialects ever get written, this won't work [23:09:12] we sometimes have more than one permanent duplicate, if the wiki has more than one version of the page [23:09:23] And if I were querying Wikidata about Berlin, I would not know to ask if there are any permanent duplicated items .... [23:11:10] Why do the Northern Frisians think they need NINE versions of the language? There is a Northern Friesian dictionary (I knew someone who worked on it), that should be the main language written form used. [23:11:29] They could have overlays for the spelling variations (which are not that much) [23:12:25] It looks like just a one-person Wiki, Murma174 [23:12:44] no idea, but it's up to them how they structure their wiki, and they're not the only ones who allow multiple versions like that [23:13:42] Interesting problem for Wikidata [23:13:50] WiseWoman: on my end, I'll just add the "duplicate item" instance_of claim to an ignore list [23:15:51] there's at least 12 others I know of, and it's also a problem for multilingual wikis (we don't have support for incubator, multilingual wikisource or beta wikiversity yet... hopefully whatever is done to support those can be used to solve the problems with wikis like the north frisian one) [23:16:34] now I'm going to bed, night! [23:16:40] Good night!