[07:06:52] ;_; https://www.wikidata.org/wiki/User:Sjoerddebruin/Weird_P31 [07:10:57] sjoerddebruin: i'm feverish, so maybe I'm wrong, could you read https://www.wikidata.org/wiki/Topic:Tdwvwpp8ldk8zj11 and give your opinion? [07:11:43] from what I understand Jura1's point of view is that if a mistake was made years ago then now it's not a mistake anymore and correcting it is changing the item [07:11:57] which is so absurd I must be missing something [07:12:36] idk [08:21:30] Harmonia_Boulot: oh, I forgot to confirm that I saw what you wrote yesterday [08:21:43] nikki: thank you! [08:22:03] nikki: we should probably start a page somewhere to list all maybe problematic descriptions? [08:23:14] I'm a bit ill at the moment though so I'm not very productive right now ;_; [08:23:19] sure [08:23:27] I'm feverish too, today [08:23:48] nikki: since the problem exist since 2013, no hurry, it can wait a week or two [08:23:59] Thiemo_WMDE: Can you have a look at the changes I did to https://gerrit.wikimedia.org/r/317127 [08:24:12] but we will correct it :) [08:24:12] also https://phabricator.wikimedia.org/T147307#2739238 [08:24:15] * nikki nods [08:24:22] that's probably of interest to you [08:30:07] also, that topic you linked above, I think you're right there. it never had any other sitelinks, it seems clear to me that the p31 is wrong, not the sitelink [08:31:25] nikki: can you say that in the discussion? [08:32:18] I think he think I always merge in given names, when no, I always search what the item was and I do things differently for each cases [08:32:31] I could understand if lots of real disambiguation sitelinks had been added to the item, but that's not the case there [08:32:40] there are cases where merging in the given name is clearly wrong and i move sitelinks then [08:33:14] I don't think we can have a unique way to treat these, actually [08:33:24] it depend [08:33:44] if the¨p31 was wrong but the item was used as such, that should come into consideration to [08:34:18] but the principal thing is that, at the end, all the items are clean, cleanly used and cleanly linked to each others [08:35:12] I won't go back to Jura1's edit when he undid my merge, but i want him to finish the job and merge into the disambiguation page if he wants that so much [08:35:27] but we can't stay with 3 items when only two are needed... [08:35:32] that's not clean at all [08:51:41] Harmonia_Boulot: done... not that I think it'll change his mind [08:52:35] nikki: well maybe he won't track all my contributions to undo that everywhere... and most importantly, maybe he won't be able to use a bot to remove all sitelinks >< [08:52:44] nikki: thank you:) [09:00:08] a while ago I came across an item that was marked as something like human settlement even though the only sitelink was for a river [09:00:28] I fixed the p31 statement, of course [09:00:38] but I guess he thinks I should have made a new item and deleted the old one :P [09:01:15] nikki: that's what I don't understand, actually [09:01:30] he added "disambiguation page" to a Vangelis album [09:01:51] should have add created a new item for the album instead of correcting his genuine mistake? [09:01:58] had* [09:02:36] Those moment where you want to add qualifiers to qualifiers. [09:02:48] How to indicate that the end date is circa? :( [09:03:05] I wish we had proper support for circa [09:04:19] (like, as part of the date/time data type itself) [09:04:20] I think we also should have a dual date datatype, also interesting for opening hours. [09:04:47] not sure what you mean by that [09:05:51] Start and end time in one datatype. [09:06:10] ah [09:06:40] But that still doesn't solve the problem I'm having... [09:18:58] Anyway. https://www.wikidata.org/wiki/Q2500611 [11:11:38] nikki: good luck with Magnus. The last thing I've heard that he doesn't want to add things to Listeria anymore... [11:11:50] :/ [11:12:38] * Harmonia_Boulot is starting to hate disambiguation items where not one sitelink is a disambiguation page [11:13:05] like, seriously, did people add P31:disambig every time they didn't feel like reading the sitelinks? [11:13:58] at least it's not a brand new feature I'm asking for, it does already support sections and it does already support sparql variables, it just doesn't support sparql variables for the sections, so maybe it wouldn't be too hard [11:15:46] Sorting on qualifiers would also be great. [11:19:05] https://grafana.wikimedia.org/dashboard/db/wikidata-datamodel-statements?panelId=4&fullscreen&from=now-1y&to=now :) [11:20:29] yes, i actually chose not to use directly Listeria on articles because I needed qualifiers sorting [11:20:51] I used Listeria on a personal subpage and then edited the code for the article [11:20:55] Also interesting: https://grafana.wikimedia.org/dashboard/db/wikidata-datamodel-statements?from=now-30d&to=now&panelId=8&fullscreen [11:21:09] and I watch my subpage for when Listeria update... [11:35:26] dennyvrandecic: do you by any chance have any of the maps or data for the maps that you generated pre in 2013 or 2014? [11:35:53] Oh right, addshore: thanks for the maps! :) [11:36:07] sjoerddebruin: no problem :) [11:36:13] I've saw some changes in South America and a specific part of Africa, right? [11:36:30] yup [11:36:56] Are you investigating the reason? :) [11:37:04] not yet! but I will :) [11:37:08] unless someone beats me ;) [11:37:28] (well probably more coords of course but it would be great to give a shoot-out to the people who did this) [11:43:47] addshore: what is the part above India? :) [11:43:54] *looks* [11:44:11] (I'm looking at https://upload.wikimedia.org/wikipedia/commons/9/96/Wikidata_Map_October_2016_Normal.png) [11:44:43] so, I have actually generate a bunch more maps, let me give you 2 links [11:45:13] http://tools.wmflabs.org/wikidata-analysis/20160404/geo2png/map_normal.png and http://tools.wmflabs.org/wikidata-analysis/20160502/geo2png/map_normal.png [11:45:21] flick between those! (thats 1 month) [11:45:50] Ah, so it's around that time period. [11:46:06] I'm working on having a map for each weekly dump, but running into some issue with downloading them form archive.org.... [11:46:24] It would be cool to have those in a slider or something [11:46:32] And something easy to compare them. [11:46:51] well, sjoerddebruin, I don't play on moving them out of that webspace now, so if you feel like making a slider ;) [11:47:38] Maybe maybe [11:48:57] so sjoerddebruin, one plan of mine was to sure up this java analysis thing, and then run it automatically after every dump [11:49:26] people could easily add their own analysis and output it straight into that webspace [11:49:38] Sounds great. And the older ones of course. [11:49:52] What is the oldest compatible dump? [11:49:56] yup, but the downloading from archive.org seems to be a bit flakey [11:50:09] So, the oldest json dump is 20141020 [11:50:41] So more than 100 folders... [11:50:45] again, in theroy we could create json dumps for data prior to that date too, and also in theroy this code can work form the xml dumps [11:51:04] * addshore doesn't have the time to lead any of this though ;) [11:52:06] sjoerddebruin: know any java? ;) [11:52:11] no no [11:52:34] When I was young, the atlas at my school contained a map with the earth at night. This reminds me of it. <3 [11:54:31] (not only because The Netherlands is the most prominent country on these) [14:19:16] Jonas_WMDE: https://www.wikidata.org/wiki/Q19798647 [15:40:19] addshore: only if it is uploaded to commons [15:40:34] okay, in that case thereis only 1 from 2013! [15:43:46] addshore: yeah, sorry none of the data has been saved anywhere I am afraid [15:43:52] wasn't there a github repo? [15:44:17] also, back then we did create the maps from the latest dump and were recreating the points in the past [15:44:18] I sill have all of the code / have rewritten it, but it only runs from the JSPOn dumps now, and they only go back to 2014! [15:44:35] dennyvrandecic: checkout http://tools.wmflabs.org/wikidata-analysis/ which is currently populating [15:44:39] they used to be working on the XML dumps though [15:44:48] ah cool :) [15:45:01] the maps still look amazing! [15:45:10] so the current code uses the Wikidata Toolkit dennyvrandecic, so in theory I should be able to make it use the xml dumps too, I just have to do some poking first I think! [15:45:59] it would be neat, but I understand time is tight :) [15:46:25] Yup, but defiantly worth a quick pass over by me, as it could be an easy 10 liner :) [15:47:28] and for reference the code is on gerrit now @ https://github.com/wikimedia/analytics-wmde-toolkit-analyzer along with a collection of other Java based processors. You can also get a built jar straight from https://github.com/wikimedia/analytics-wmde-toolkit-analyzer-build [15:47:39] the actual image generation is still using the same python script ;) [15:51:15] oh dear :D [15:51:35] do I really want to read my four year old python code? :D [16:02:07] dennyvrandecic: even worse, its your 4 year old python code that has been mutilated by me! [16:03:00] I'd be surprised if you made it any worse though :D [16:06:14] hah, well, python is not my tidiest programming language dennyvrandecic ;) (But I guess thats why I rewrote a bunch of it in Java) :D [16:06:23] do you have a higher resolution one of this: https://upload.wikimedia.org/wikipedia/commons/9/96/Wikidata_Map_October_2016_Normal.png [16:07:25] tidiness in programming is more correlated to the author than to the language in my experience, addshore... ;) [16:08:01] dennyvrandecic: https://commons.wikimedia.org/w/index.php?search=Wikidata+Map+October+2016&title=Special:Search&fulltext=Search&searchToken=a9zngu5tar8847pnp09jw7ql6 [16:08:36] you have Big Huge and Enormous ;) [16:09:01] Wow, Google processed my report pretty fast. Albert Bessemans is dead for real, though he passed away in 1973. [16:09:21] I am just exploring enormous, wow [16:09:45] sjoerddebruin: you mean you feedbacked a knowledge graph result of a person who wasn't marked dead? [16:09:57] Yup. [16:10:13] Better response than Maps, that's for sure. [16:10:33] sjoerddebruin: oh,wow, I usually here it the other way around [16:11:47] sjoerddebruin: did you get any feedback like "thanks" or "it is done now" or did you just check it again? [16:11:53] The latter. [16:12:04] And it seems like my Maps issue is fixed too, though. [16:12:15] thanks for both reports! :) [16:12:36] Not sure though... [16:13:09] I hope some day the Knowledge Graph would pick more of my added information. :3 [16:14:28] Hm, nope I was confused. The Maps issue still exists. [16:21:20] Is this normal display? https://www.dropbox.com/s/nxip1wiehm421xt/Schermafdruk%202016-10-25%2018.21.13.png?dl=0 [16:22:17] The display on https://www.wikidata.org/wiki/Special:Contributions/Sjoerddebruin differs from https://www.wikidata.org/w/index.php?title=Q6721195&action=history... [16:23:22] sjoerddebruin: Special:Contributions has troubles with edits done in the same second [16:23:23] ignore [16:23:29] Ah, okay. [16:33:55] Could someone have a look at the changes I did to https://gerrit.wikimedia.org/r/317138 and +1 maybe? [17:18:58] SMalyshev: uh, what should the value of the sparqlEndpoint setting look like? [17:19:09] where would i find out the correct url to use with wikidata? [17:22:51] ah, https://query.wikidata.org/bigdata/namespace/wdq/sparql?query={SPARQL} [17:22:55] well hidden :) [18:18:25] DanielK_WMDE: yes [18:19:07] DanielK_WMDE: or simpler: $wgWBRepoSettings['sparqlEndpoint'] = 'https://query.wikidata.org/sparql'; [18:19:28] SMalyshev: Sometimes sparql is like black magic. Do you know if sparql has something like SQL EXPLAIN? [18:19:59] multichill: well, sparql has explain - or, more precisely, blazegraph has explain [18:20:14] I feel a but coming up [18:20:16] the question whether it makes matters more clear, is much harder :) [18:20:40] it will tell you the query plan, however understanding it is not really easier :) [18:20:53] nikki and I have been battling queries that time out, but if you shift them around, complete in time [18:21:08] SMalyshev: is that the plan after optimizations have been applied? (reordering?) [18:21:11] that's completely possible, optimizer is not perfect [18:21:23] that could be useful to understand those issues [18:21:24] WikidataFacts: it provides both [18:21:29] even better [18:22:12] I think if you add explain=true to REST API call it should give you explain [18:22:44] http://tinyurl.com/hlcel8u ? [18:23:22] e.g. this: http://tinyurl.com/h7v82qw [18:23:44] Ah, nice [18:23:49] So for example see https://www.wikidata.org/wiki/Wikidata:SPARQL_query_service/queries/examples#Authority_control_properties_usage_for_paintings [18:23:55] And the next one [18:24:22] Both basically the same query, but for a different domain. Second one only completes if you make it a subquery [18:24:31] First one doesn't complete if you make it a subquery [18:25:43] as I say, optimizer is not perfect... I've asked to make query hints that allow manual ordering or some kind of hints to give to optimizer, since we sometimes know the structure of data better than it does. [18:26:49] I try to have query with the least lines as first line and than the others [18:26:59] But if the optimizer shuffles that around, that won't work [18:27:09] I have a feeling maybe I need to add some kind of explain button to the gui... shouldn't be that hard [18:27:28] multichill: you can disable optimizer if it messes up things [18:27:49] hint:Query hint:optimizer "None" . [18:27:58] see: https://wiki.blazegraph.com/wiki/index.php/QueryHints [18:29:37] that runOnce thing sounds useful [18:29:49] * WikidataFacts looks up NamedSubquery [18:30:38] wow, this sounds like what you’d want subqueries to behave like: https://wiki.blazegraph.com/wiki/index.php/NamedSubquery [18:32:34] WikidataFacts: That looks like my VALUES work around ;-) [18:32:52] exactly, “run the query once and then insert the results” [18:36:59] WikidataFacts: http://tinyurl.com/j2w4crr [18:43:37] Hey, just read that Europeana has SPARQL too http://labs.europeana.eu/api/linked-open-data-SPARQL-endpoint [19:13:03] OMDB is quite interesting, see http://www.omdbapi.com/?i=tt0096697&Season=27 and http://www.omdbapi.com/?i=tt0701047&r=json [19:14:28] hey sjoerddebruin [19:14:33] hey multichill [20:05:26] SPARQL question: I want a list of statements (and corresponding items) that have P248: Q20022913 as their reference. What would be the query for that? [20:10:31] hare: http://tinyurl.com/h8xp9hf [20:11:22] Now what about that, but for reference URL beginning with "http://www.cdc.gov/niosh/" [20:11:42] I suspect that’s going to be too expensive… let’s see [20:11:59] I mean I've done it before, but then I lost the query D: [20:13:45] Did I do this right? https://www.wikidata.org/wiki/Q27556095 [20:13:59] hare: do you mean, have P248 Q2... *and* reference url beginning with ... as references? [20:14:09] because searching just for reference URL gives me timeout [20:14:10] no, one or the other [20:14:35] do you have any other restrictions I could add? [20:15:00] Not really, unfortunately. Also I don't really need the full statement spelled out; just the hash will do. [20:15:34] well searching through all reference URLs everywhere is just going to be expensive, I don’t see a way around that :/ [20:16:24] LIMIT doesn’t help either [20:17:32] wait, no, searching just for the statement might work [20:17:41] That's what we did last time that worked ;) [20:17:56] http://tinyurl.com/h5zorma – 24340 results, good lord [20:18:03] I’m surprised my browser handled that so well [20:18:09] 25 seconds too, cutting it close [20:20:02] So, let's combine that all together. A count of statements with references of P248:Q20022913, P248:Q17144694, or a reference URL that begins with that. (If a statement has more than one of those things it only counts once.) [20:20:24] I do appreciate your help. I managed to come up with this query about a month ago and then proceeded to lose it. :( [20:22:11] 26320 of them: http://tinyurl.com/zek8mlk [20:22:26] Yay! Thank you [20:22:35] you’re welcome :) [20:23:01] Now, I am going to save this so I don't lose it!!!! [20:23:13] just search the IRC logs next time :P [20:30:00] WikidataFacts: http://tinyurl.com/h4bjvub << and this is for, instead of P248:Q20022913, the statement is "stated in" any item with a P2880? [20:30:52] yup, that should wrok [20:30:54] *work [20:31:10] Alright, thank you [20:31:52] month over month i'm down 560 statements. i wonder if it's because the two queries are slightly different, or statements were removed, or what [20:32:01] It may have something to do with the work I've been doing with chemicals [21:02:49] "curses, wikidata locates the astronomical observatory of trieste in slovenia instead of italy in the infobox map ! fix it, this is unacceptable, you can't trust wikidata once again !" - "uh, actually, trieste is indeed in italy, but on a band of land that's 10km wide and 50km long on what would otherwise be the slovenian waterfront, zoom on the map to see what's happening there" - "oh, carry on then" (someone on frwiki's village pump) [21:03:20] ugh [21:26:29] Alphos: :D [21:32:51] dennyvrandecic it's not funny :-( ok, maybe a little, but it shows people are rejecting wikidata simply because they don't like it [21:33:15] Alphos: well, but that is the funny thing :) [21:33:35] give me the strength to change what can be changed and laugh about the things that cannot be changed [21:33:50] rejecting wikidata just because it is wikidata cannot be changed [21:34:53] so we can make it as good as it gets in order to become as valuable as it can be, and then it does not matter whether you like it or not [21:35:08] and working on making it good, well, that's what we can do :) there's something we can change [21:35:13] i tend to react poorly with people rejecting things offhand when they have no idea of what the things they're rejecting are [21:36:25] I guess that makes you rather unhappy with people in general :) [21:39:31] less and less [21:40:06] that is good :) less unhappy sounds like a good place to be [21:43:06] XD [21:43:34] <3