[00:21:24] yurik: yes? [00:21:57] SMalyshev, what do you think about https://phabricator.wikimedia.org/T181319 [00:22:51] yurik: you can not put literal in subject position in SPARQL [00:23:08] otherwise it can work, technically [00:23:16] SMalyshev, sure, we can use some prefix instead [00:23:27] its a url encoding, so anything should work in theory [00:23:42] not sure it will perform though [00:23:43] as no indexing, obviously [00:23:45] so the question is why [00:24:02] i wonder about the triples - the CSV is not a tripple store [00:24:22] to be able to combine public data sources with WD [00:25:13] we are amassing tons of tabular data in the Commons, plus there are many public data sources like the example. To be able to join them would be a very powerful use [00:25:48] combine how? [00:25:55] note - no indexing [00:26:12] of course - this would only work for a relatively small datasets [00:26:53] for example - you can take a long list of stocks as a csv page, and find the companies that own them [00:27:01] well there's nothing impossible in this technically as far as I can see [00:27:13] sparql result is basically a table, so is csv [00:27:31] how would FILTER inside the SERVICE function? [00:27:51] would it work like a row filter? [00:28:01] or is it something that the SERVICE provider needs to implement? [00:28:33] also, does service provider need to specify some wildcard of all the URLs it would handle? Or can there be a more generic approach [00:38:05] FILTER won't work inside SERVICE [00:38:21] unless your service provider implements SPARQL of course [00:38:38] interpreting inside of the SERVICE is entirely up to service provider [00:39:07] also, providers are identified by service URL, so you can't put CVS there [00:49:56] yurik: however after result set has been retrieved, you can apply filters. But no indexing and no filters inside, unless you implement it outside, so if you have 1000-line CSV, and you need one row, it'll download all lines, scan them one by one and then select one and throw out the rest [00:50:09] not very efficient... but that's why people have databases :) [01:13:54] CALL 415-349-5666 L0DE RADIO HOUR [04:14:22] is it me or is it petscan, however, for enWP I add one category "Bowls players at the 2014 Commonwealth Games‎ and put in 0 or 1 levels and get zero hits [11:20:46] Hi! I'm setting up a Wikibase site outside Wikidata, and now is looking for some way to query (e.g. an SPARQL API endpoint like query.wikidata.org/sparql) in the Site. My question is: currently, is there any existing (under development or not) extension or something else that can help me to do it? [11:20:59] monitor domain upayogee.com spam [11:21:10] grr [11:22:23] monitor domain karnopedia.com CoI spam, non-authoritative [11:49:20] cxuesong: there’s no extension for that, but you can install WDQS using the instructions here: https://www.mediawiki.org/wiki/Wikidata_query_service/User_Manual#Standalone_service [11:58:16] cxuesong: https://github.com/wmde/wikibase-docker ? [12:08:25] Thanks, Lucas & multichill! I will take a look at them. [13:28:43] Another question: Currently I have no means to refer to a foreign entity (e.g. refer to an entity on Wikidata from my site), if my site and foreign one does not share the same DB. Is that right? [13:31:06] cxuesong: correct, as far as I know [13:32:51] Thanks for your confirmation. I've already started populating my repos. [20:33:06] Ughhhhhh. https://www.wikidata.org/w/index.php?title=Q5361934&action=history [20:48:19] Why do people rename main namespace articles to categories and merge them on Wikidata? :( [20:50:34] I think the answer to every question ever that starts with "why do people" is "because people, sigh" [20:50:38] Articles about settlements renaming to categories about districts... [20:51:25] * reosarevok earlier split some random Estonian church item from the concept of "peace churches". I still can't understand how they ended up together (I didn't dare to check history) [20:52:00] Cleaning up mandatory constraints piles me up with anger, but nobody is doing the task at the same time. [20:53:04] ... https://www.wikidata.org/w/index.php?title=Q11963180&action=history [20:53:41] The item seems to be describing it correctly though. [20:53:43] Weird stuff. [20:54:05] Is there a list of stuff with weird constraints that needs checking? [20:54:15] I know the per-property constraints, but nothing else yet :) [20:54:33] Weird constraints? Probably via API somewhere. [20:54:56] eh, I meant "weird constraint failures" as in this "both person and company" issue here [20:55:53] * reosarevok is not great at explaining himself today :) [20:55:57] https://www.wikidata.org/wiki/Wikidata:Database_reports/Constraint_violations/Mandatory_constraints/Violations [20:57:09] Oh god that's a lot. Guess I can take a look at fixing a few every once in a while at least, maybe it'll help [20:57:22] Thanks! [21:00:49] So for example, for https://www.wikidata.org/wiki/Q7913799 (a duo of males) would the right choice be to just remove "sex or gender"? (and probably "entertainer", "singer", "comedian" too I assume) [21:01:07] Yeah, probably. [21:01:25] IMDB id is about one of them. [21:01:38] So you need to check and create items for both of them if you want to fix it correctly. [21:03:56] Gotcha [21:04:15] And I see "member" is an alias of "has part" [21:04:45] Yes. [21:05:05] Like https://www.wikidata.org/wiki/Q484918 [21:05:51] heh, and I see this constraints gadget I set up tells me "Art Garfunkel should also have the inverse statement part of Simon & Garfunkel." [21:05:56] Yay, rabbit hole :p [21:06:49] * reosarevok wonders why the gadget doesn't have a "yes, looks good, make it happen" button [21:07:04] There are plans for that, at least for trusted users. [21:08:45] Neat [21:08:57] Yeah, I'd understand if I didn't qualify for that yet :) [21:09:31] https://phabricator.wikimedia.org/T167700 [21:12:24] :) [21:12:26] Sounds good to me [22:44:28] "image: sex or gender man" [22:44:29] wat [22:45:37] Heh, not sure what to do with https://www.wikidata.org/wiki/Q16738939 [22:46:13] heh. "author: author" [22:47:56] heh, and quote used as "famous quote by this person" [22:53:42] https://www.wikidata.org/wiki/Q121594 should maybe be considered a position or occupation? (not just a title) [22:55:26] It seems it clashes now with stuff like "female form of", which seems like a reasonable thing for it to have [22:58:30] I'm moving https://www.wikidata.org/wiki/Q42405008 from "author" to "occupation" for now, but I'm not sure if it should be an item and if so how to link it, it's not really an occupation either... [22:58:48] Possition held? [23:02:39] * reosarevok would just remove them all and if anything use the people as "author" for a specific per-article item, but not sure [23:04:45] If we clean up mandatory violations, we might find things like https://www.wikidata.org/w/index.php?title=Q6180996&action=history faster [23:05:21] https://www.wikidata.org/wiki/Q121594 is a mess. [23:08:42] Looks like it. And yeah, relatively empty reports are always useful [23:08:55] To be fair most of this is a huge ball of mess [23:09:20] Only stable constraints are supposed to be marked as mandatory. [23:09:26] (or those who are easy fixable) [23:09:58] Most of the times, "unique value" isn't both of those. [23:10:23] Well, "person can't be authored" which I'm doing now is easily fixable, all these seem to basically require "remove author, do nothing" [23:10:33] Yeah, pretty confusing property. [23:10:47] English description is clear, but I'm betting the Greek one is less so [23:14:02] Well, or this one person who added a ton of entries didn't read it :p [23:14:28] Ooh, a male poem :D [23:17:30] Sigh. https://www.wikidata.org/w/index.php?title=Q30905721&offset=&limit=500&action=history [23:19:05] spam? [23:19:48] heh. https://www.wikidata.org/wiki/Q1068491 and https://www.wikidata.org/wiki/Q3306444 are "characters", with a set author. what's the right property for that? [23:20:15] P170, I guess. [23:20:42] Ok, well they have those, so remove author and done [23:20:46] Nice [23:21:23] https://www.wikidata.org/wiki/Q852034 - what's the way of saying "it's always female" without using p21? [23:21:58] https://www.wikidata.org/wiki/Property:P1552? [23:21:58] And then https://www.wikidata.org/wiki/Q467 as value? [23:24:56] Sigh, https://www.wikidata.org/wiki/Q1632277 is both character and series. No time to fix it now (too late) but I guess if I fix the easiest ones this will be easier to find :p [23:29:58] We cleaned up for 60k already. ;) [23:39:37] Well I said easier, not easy! ;p [23:43:57] https://www.wikidata.org/wiki/Q8834659 "instance of: LOL". sigh :p [23:44:24] Done on 3 october... [23:47:37] 27% of our constraints are mandatory. [23:50:48] Too many? [23:50:54] Nah, could be much more. [23:53:16] Maybe, but then they should be much more obvious to the adding users, or we'll end up with a much bigger mess [23:53:22] * reosarevok needs to get some sleep now though :)