[00:10:50] PinkAmpersand: I need someone to translate "Hello, I'm just visiting (I don't speak French). Can someone please take a look at http://fr.wikipedia.org/wiki/Elbow_Ledge and see if it needs deleting? " into French for me. Can you pull that off? [00:12:34] it'll depend on their CSD criteria [00:12:54] Sven_Manguard: kk 1 sec [00:13:06] never mind [00:13:11] oh [00:13:11] fabriceferrer is on it [00:13:22] o/ [00:13:30] I posted that (in English) over in pedia-fr [00:15:40] I hate those merge conflicts.. ;/ [01:00:49] legoktm: I have a bot idea [01:00:56] Sure [01:00:59] What is it? [01:01:02] gimme a sec to write it [01:01:13] ok :P [01:01:13] So take a look at http://www.wikidata.org/wiki/Q134091 [01:01:22] wait [01:01:24] no [01:01:28] http://www.wikidata.org/wiki/Q4987277 [01:01:30] that [01:01:50] ok [01:01:52] if we have the genus filled out in an item, a bot could easily fill out the other taxon data [01:01:58] hm [01:02:07] where does it get the data from? [01:02:12] since all items in the same genus are in the same family [01:02:15] oh https://www.wikidata.org/wiki/Q134091 [01:02:16] i see [01:02:24] just upwards recursion. [01:02:31] right [01:02:51] we might even be able to find some proper mapping somewhere like EOL that we could borrow [01:02:55] EOL? [01:03:00] Encyclopedia of Life [01:03:19] see the little box on the right in http://eol.org/pages/16863112/overview [01:03:53] err.. no wait [01:04:12] ooh [01:04:13] API [01:04:17] yeah [01:04:19] i think [01:04:21] this can be done [01:04:44] err... yeah, actually I think that those tables are right [01:05:05] heh [01:05:06] http://eol.org/data_objects/18598368 [01:05:30] question [01:05:39] looking at [01:05:40] https://en.wikipedia.org/wiki/Bulbophyllum_acanthoglossum [01:05:47] there is no link to EOL? [01:05:58] no, there is not [01:06:01] EOL is not our project [01:06:04] hmm [01:06:13] Wikispecies? I forget what that project does [01:06:27] it would be nice if there was an easy way to go from item-->EOL [01:06:31] without just a name search [01:06:41] it might be item-->enwiki-->wikispecies-->EOL [01:06:44] i wouldnt care [01:06:46] https://species.wikimedia.org/wiki/Bulbophyllum [01:06:53] would that work? [01:07:05] it has it up at the top [01:07:20] yeah that would [01:07:23] plus copyright would be better [01:07:29] yeah [01:07:35] admittedly Wikispecies isn't the most well developed project [01:07:46] Wikispecies shall be absorbed by wikidata [01:07:48] >:D [01:07:49] EOL has a lot of bot gathered data, so it's more populated [01:07:56] and now we're just importing it [01:07:56] yeah [01:08:01] brb in 10 min [01:08:04] Ajraddatz: honestly I don't see why not [01:08:20] Wikidata could do it eventually [01:08:30] because some of the people who are on Wikispecies don't want to give up their corner of the internet [01:10:03] * Jasper_Deng pokes Jeblad_WMDE [01:10:42] hm? [01:12:31] is there a mechanism in wikidata for importing ontologies / structured vocabularies? [01:12:51] * Jasper_Deng just PM'd Jeblad_WMDE [01:12:53] (/ will there be?) [01:13:53] for example, if wikidata were to support the idea of classes/types more fully, would there be a way to import third-party ontologies like GND (http://d-nb.info/standards/elementset/gnd)? [01:13:55] New proposal: Non-autoconfirmed need to get a popup whenever they try to remove a link from an item. [01:15:33] Ajraddatz: warning? [01:15:38] we already have an abusefilter that tracks that [01:15:41] yeah [01:15:49] maybe the abusefilter should actively warn them then [01:15:55] because 99% of the time, the edit is bad [01:15:58] https://www.wikidata.org/wiki/Special:AbuseFilter/10 [01:16:28] How do af warnings work on items? [01:16:30] There are no import of ontologies [01:16:32] not sure [01:16:33] test it? [01:16:40] There will probably be none either [01:16:48] is abusefilter on the test wiki? [01:16:56] no [01:16:56] just [01:16:56] or do I get to mess up the main wiki :D? [01:16:57] WOOT! [01:17:03] set up a filter with user_name="Ajraddatz" [01:17:08] hehe yeah :p [01:17:24] whoa, that works? [01:17:32] I've always used user_name rlike "Ajraddatz" [01:17:43] but I suck at everything so I suppose it's a moot point :3 [01:17:48] rlike is regex [01:17:51] Sven_Manguard: i think i figured it out [01:17:52] yeah I know [01:18:00] I didn't realize there was an exact match command [01:18:04] one question [01:18:06] or whatever you call it [01:18:16] ye? [01:18:23] for these scientific names, can i safely assume there will be no overlap? [01:18:37] ie if i search " Suifunema mackenziei" there will be 0 conflicts? [01:18:40] Sven_Manguard: ^ [01:18:41] theoretically there shouldn't be [01:18:48] theoretically? :P [01:18:51] if there were, I'd be utterly shocked [01:19:01] most likely an interwiki screwup [01:19:08] there're designed to be unique [01:19:12] Jasper_Deng_away: I have not been involved in development of that function [01:19:14] they're useless otherwise [01:19:19] ok [01:19:33] lemme check what EOL's copyright policy is [01:19:33] Jasper_Deng_away: What I've heard is that it does not work for enhanched wtched changes, but should work for the normal changes [01:19:33] but I'm not a biologist, so lemme check [01:20:02] warning does not show up on item creation [01:20:15] :o [01:20:19] hoo: ^ [01:20:26] AbuseFilter is not something you should change if you don't know what you are doing [01:20:28] Cool. Warnings show up like an editconflict box [01:20:39] I do know what I'm doing, though. [01:21:01] Yes, legoktm according to [[Binomial nomenclature]] they are unique [01:21:01] 10[1] 10https://wikidata.org/wiki/Binomial_nomenclature [01:21:03] My test filter works perfectly. [01:21:06] ok [01:21:08] There have been incidents where the total influx of changes to wmf-projects have dropped 60% due to misconfigured filters [01:21:18] :x [01:21:27] Its just username=="blah" [01:21:29] so its safe [01:21:33] What are unique? [01:21:42] Emw: Binomial nomenclature names [01:21:47] legoktm: What are you talking about? [01:21:48] like Bulbophyllum acanthoglossum [01:21:54] Can you summarize it? [01:21:56] I really would like access to AbuseFilter being turnde off for ordinary admins as I think the possibillity for screwups are to big [01:21:56] hoo: [07:20:03 PM] warning does not show up on item creation [01:22:05] So yeah, the warning works on article edit (warning notice in the same manner as an edit conflict), but not upon item creation. [01:22:25] Jeblad_WMDE, probably a good idea. [01:22:26] hoo: its a simple filter with user_name="Ajraddatz" and warning set [01:22:27] legoktm: Ewk :/ [01:22:34] For well-defined species, I think that's a good bet [01:22:43] [20:00] Sven_Manguard legoktm: I have a bot idea [01:22:45] [20:01] Sven_Manguard since all items in the same genus are in the same family [01:22:46] [20:01] Sven_Manguard since all items in the same genus are in the same family [01:22:48] [20:02] legoktm oh https://www.wikidata.org/wiki/Q134091 [01:22:49] [20:02] legoktm i see [01:22:50] aaand filter disabled [01:22:51] [20:02] legoktm just upwards recursion. [01:22:53] back to homework [01:22:57] hoo ^ [01:23:06] legoktm: Is it doing the change silently or going back the form for no appearent reason? [01:23:13] hoo: idk ask Ajraddatz [01:23:14] err [01:23:15] I would also like a thorough review of all filters that goes into production [01:23:31] Some of the "mistakes" are plainly horrible [01:23:36] Jeblad_WMDE: I would like that for gadgets... [01:23:38] [20:00] Sven_Manguard legoktm: I have a bot idea [01:23:40] [20:01] Sven_Manguard if we have the genus filled out in an item, a bot could easily fill out the other taxon data* [01:23:41] [20:01] Sven_Manguard since all items in the same genus are in the same family [01:24:01] hoo, with item creation no warning notice appears anywhere. It just goes through with page creation as normal. [01:24:32] Ajraddatz: Mhm, can you open a bug for that, please? [01:24:43] Sven_Manguard: I think better Wikidata support for ontologies would fix that [01:24:45] cc me please [01:24:51] Probably the non-JS special pages are affected as well... [01:25:07] blarg, in a bit [01:25:10] I'm a bit busy now :s [01:25:13] got a midterm tomorrow [01:25:22] stupid university [01:25:31] e.g., you fill out an item as being a given type, then Wikidata fills in the inherited properties [01:25:33] why can't I just work at McDonalds? [01:25:34] ok [01:25:40] Ajraddatz: if only it was that easy [01:25:42] ill file it [01:25:44] thanks [01:26:12] Traditional biological taxonomy is a good example, but there are many others [01:27:25] Emw: but in the mean time, having a bot fill out those properties makes sense [01:28:06] Ajraddatz: did it not get logged in https://www.wikidata.org/w/index.php?title=Special:AbuseLog&wpSearchUser=Ajraddatz ? [01:28:15] emw, try to define a working ontology for 25 mill entries that are complete... [01:28:30] Jeblad: how many infoboxes are there? [01:28:38] Emw: about 5000 [01:28:48] Or rather, just read Dennys blogpost [01:28:55] sorry, how many infobox instances? [01:29:02] the page edits did [01:29:06] but not the creation [01:29:19] Emw: across all wikis? 10 million or so, i guess [01:29:22] ok [01:29:44] Jeblad_WMDE: oh, denny bogged? link? [01:29:54] filed https://bugzilla.wikimedia.org/show_bug.cgi?id=45537, cc'd both [01:30:10] ty [01:30:12] DanielK_WMDE_: i think the success of infoboxes indicates that ontology support would be very helpful in wikidata [01:30:20] DanielK_WMDE_: WMDEs blog [01:30:43] I like his examples,.. They seems stupid but he has done his homework [01:30:47] Emw: i don't see how these things are related. and the term "ontology" is so heavily abused, i'm not sure what you mean. [01:31:08] Emw: i agree that some kind of schema or prototype support would be nice (though others in the team disagree) [01:31:17] that does *not* imply anythig about ontologies. [01:31:17] e.g., supporting property inheritance in items that specify a type [01:31:42] Emw: would be nice, and also introduces no end of problems [01:31:47] Ontology for biological species are probably just the latin name,.. its not really an ontology [01:32:34] Emw: to make this feasible, the inherited properties need to be materialized in the (sub-)item. Otherwise, things get, err, "fun". [01:32:48] whats the name of inheritance when it comes to species.. nah, I go back to the rebase from Hell.. [01:32:51] and this can be done by bot easily enough [01:33:03] DanielK_WMDE_: what does "materialized" mean here? [01:33:10] Emw: copied [01:33:23] heh, enjoy your rebase :P [01:33:51] Why would they need to be copied? Couldn't they be looked up upon demand? [01:34:18] Somebody wants some code reorganized in separate patchsets, I will not say whom as he is in the channel.. ;p [01:34:41] Emw: recursive lookups are slow. and we have no guarantee ther eare no circular dependencies. also, your item would change without you seeing it in the history. and the system then needs to track changes recursively, so it knows when to notify client wikis about changes [01:34:43] etc etc [01:34:45] * Jeblad_WMDE kicks DanielK_WMDE_ in the butt [01:35:07] Jeblad_WMDE: i'll do the rebase if you like [01:35:09] gimme a link [01:35:48] Emw: we *did* think about this, and this has been tried. it's not easy. it's possible to some extens, but pretty hard to do efficiently for large amounts of data [01:36:27] Emw: wikidata is designed to be flexible and wiki-like. introducing too much automatisms takes away flexibility, and adds a lot of technical dependencies and difficulties. [01:36:52] Sven_Manguard: Is there a master cat/template on enwiki that i can search to get a list of names of all the biological things? [01:37:23] DanielK_WMDE_: to what extent is it possible to implement support for inheritance? [01:37:26] hmm [01:38:15] Emw: in wikidata? i could imagine something where you'd say "create a new item like that one", and then you pick what to copy from the original. [01:38:26] i would not try to go beyond that [01:38:53] Emw: that could be done with a gadget, btw. no need for server side support [01:38:53] legoktm: I don't suppose http://en.wikipedia.org/wiki/List_of_mammal_genera would help? [01:39:32] wait [01:39:36] hi Emw [01:39:43] http://en.wikipedia.org/wiki/Category:Taxonomic_categories [01:39:45] hi aude! [01:39:58] coming to our meetup tomorrow? :) [01:40:04] Sven_Manguard: can i just use https://en.wikipedia.org/wiki/Template:Taxobox ? [01:40:15] legoktm: I have no idea [01:40:19] :P [01:40:21] i can't program, remember [01:40:54] if it has the information you need, then sure [01:40:57] ill look into it [01:41:04] if i can, this means i dont have to write separate code [01:41:12] I imagine that inheritance will grow organically (as is being discussed now with taxonomies), but that it will be less robust and more server-resource straining than if it were given decent support in Wikidata [01:41:25] i can just use my template parsing code [01:42:22] e.g., a patchwork of bots will be making these edits, updating properties in items that inherit from other items, which Wikidata would be doing anyways [01:43:27] for others interested in this, Denny's blog post is relevant: http://blog.wikimedia.de/2013/02/22/restricting-the-world/ [01:44:25] i think it would be challenging to do inheritance in the code [01:45:12] e.g. United States is in both North America and Asia (e.g. Guam) [01:45:21] so, should New York City be in Asia? :) [01:46:56] * legoktm would like to go to Asia next weekend :P [01:47:01] heh [01:48:23] aude: i think New York City could be determined to be in North America its containing geography [01:48:35] by evaluating its* [01:49:22] maybe but i think there will be lots of edge cases like that [01:49:25] e.g. NYC -> ... -> NY -> North America, perhaps [01:49:51] what is the proportion of edge cases to non-edge cases with inheritance? [01:50:04] a bot could have some amount of human guidance [01:50:21] * aude has no idea but significant i think [01:50:45] sure, a donkey was a Roman emperor, but all the all others were human [01:53:58] all the other donkeys was humans? [01:54:11] * Jeblad_WMDE scratches head [01:55:09] Some of the old german states had laws that also covered animals [01:55:55] It was not uncommon whit public spanking of animals that created public disturbances [01:56:12] I even have a book that documents some of the cases [01:56:44] Sometimes reality is more weird than the fantacy [01:56:52] err, apparently there wasn't a donkey Roman emperor, but my point is that i think the overwhelming majority of items in wikidata would not be edge cases for inheritance [01:59:11] It is better to create methods to identify common cases, and also errors, than to create rigid systems where everything must be defined [02:05:27] Jeblad_WMDE: I don't think inheritance support in Wikidata would imply that everything must be defined (as in, have a defined type, and fill in all of the properties comprising that type) [02:08:33] The problem of finding possible entries for a property is much the same as finding likely products for a prospective customer. That is a more flexible approach than creating a fixed list,.. but I could be wrong. [02:25:33] Sven_Manguard: zombie!!!!! https://en.wikipedia.org/w/index.php?diff=541114319&oldid=540545599 [02:25:39] lol [02:25:51] * PinkAmpersand blocks Sven for violating the laws of nature [02:26:15] you aren't an admin on Wikipedia, and I have more powerful friends over there than you do [02:26:20] Emw: there was a horse Roman senator... [02:26:24] and if you block me on Wikidata... well... [02:26:26] heh [02:26:36] haha ;) [02:26:36] Sven_Manguard: you left fast [02:26:43] he? [02:26:53] Sven_Manguard: Like Jimbo? :P [02:27:07] Hazard-SJ: I don't count him as a friend [02:27:15] actually we've traded nasty emails more than once [02:27:32] * PinkAmpersand 's favorite Wikipedia feud is Jimbo vs. Fram [02:27:38] legoktm: I've started recoding my bot for sitelinks :D [02:27:47] yay! [02:27:49] it should serve as an indication of my general views towards him that I *brag* about trading nasty emails with Jimbo [02:28:57] legoktm: I added a new part to it too, in fact. Not sure if anything works as yet, I haven't tested anything, but I'd need another bot request for approval for the new section [02:29:07] what does it do? [02:29:28] i dont know why youd need a new bot approval [02:29:36] Sven_Manguard: seriously, though, list yourself on the Wikidatans list by all means, but if you're not gonna be away for extended periods of time, maybe add like a (less active than usual) or something? [02:29:45] meh [02:30:02] you clearly don't understand the relationship I have with inactivity [02:30:10] I used to put 40 hours a week into Wikipedia [02:30:14] :x [02:30:16] did it for about a year and a half [02:30:20] sven is inactive? [02:30:37] me being inactive means that I only check it once every day or so [02:30:51] and don't edit for several hours [02:30:58] ok :) [02:31:19] I mean, in all likelyhood I will have 0 edits next week because I'll be doing RL work [02:31:33] but that doesn't mean I won't still check to make sure nothing is on fire [02:32:52] and if someone emails me at my wikipedia email account, I know within 15 minutes [02:33:22] well thats slow. mine just forwards straight into my main inbox. [02:34:00] yeah. i once pinged a guy who'd been inactive for months, because i thought he might be someone i knew from school, and he replied in like 2 days. [02:35:07] legoktm: the 15 minutes is because my phone's battery lasts way longer if it only checks for mail every 15 [02:35:12] ah [02:35:19] pull, not push [02:35:47] I didn't use either [02:35:51] not a techie person [02:36:15] hey, lego, can you add <(n)?> after , , and on Filter 8? I'm afraid I'll fuck something up again [02:36:23] um [02:36:43] oh ok [02:36:44] i see [02:39:11] done [02:41:53] legoktm: Your empty items bot is still down? [02:41:59] yeah sorry [02:42:03] ive been very busy lately [02:42:39] legoktm: The new part checks for moved pages and updates the sitelinks on Wikidata where possible [02:42:47] oh [02:42:55] Bene16 is taking care of that [02:43:11] https://www.wikidata.org/wiki/Wikidata:RfBA#BetaBot [02:43:15] Beta16* [02:44:42] :O [02:48:02] legoktm: I'm re-enabling reporting (without any additional fixes) [02:48:09] ok [02:48:10] sounds good [02:59:41] legoktm: How do I get SVN access for Pywikipedia if there is no consensus? :S [02:59:56] have you submitted patches? [03:00:14] No [03:00:43] well do that [03:02:09] * Hazard-SJ decides to do two simple changes that have been annoying him, then get back to his coding [03:03:12] just stick the patch on sourceforge and send me the link and ill commit it [03:05:51] OK [03:05:59] legoktm: just added "filipino" to edit filter 8. mind checking it to make sure I didn't b0rk the wiki this time? [03:06:08] how about [03:06:12] you use the test functions [03:06:15] BEFORE saving? [03:06:40] because [03:06:41] if you had [03:06:52] you would have noticed your regex was horribly broken [03:07:15] oh [03:07:16] wait [03:07:17] ugh [03:07:23] hold on [03:07:33] * legoktm takes that back [03:07:35] it was fine [03:07:41] sorry :x [03:07:47] but please use test functions BEFORE saving [03:07:51] phew. didn't think i could do much worse than a markup error [03:08:15] https://www.wikidata.org/wiki/Special:AbuseFilter/test [03:08:27] hit [03:08:29] load 8 [03:08:31] oh yeah that's a good idea [03:08:33] find the edit you want to test again [03:08:35] against* [03:22:43] oh cool [03:22:45] http://legalblogwatch.typepad.com/legal_blog_watch/2013/02/alberta-court-concludes-that-access-to-internet-is-fundamental-part-of-right-to-counsel.html [03:23:00] heh [03:23:09] * legoktm wonders who Sven_Manguard is getting sued by [03:23:23] That's not me, I'm not Canadian [03:23:32] > He also testified that because he believed ("based on Hollywood movies") that he was only entitled to one telephone call, he did not request any further opportunities to get legal advice at the time. [03:23:40] Oh right. [03:25:17] legoktm: https://sourceforge.net/tracker/?func=detail&aid=3606291&group_id=93107&atid=603140 [03:26:48] looking [03:28:56] ahh [03:28:57] sorry [03:29:00] my finger slipped [03:29:03] https://www.mediawiki.org/wiki/Special:Code/pywikipedia/11137 [03:29:20] Hazard-SJ: ^ [03:48:32] legoktm: Thanks, and it's ok [03:48:53] also [03:49:04] what do you think of [03:49:04] https://www.mediawiki.org/wiki/Manual:Pywikipediabot/Wikidata/Rewrite_proposal [03:49:09] for rewrite branch [03:49:39] would let you do stuff like [03:49:48] prop=Property(site, 'P129') [03:49:57] prop.target='Q125' [03:50:16] source=Property(site,'P3') [03:50:24] source.target='Q1' [03:50:27] prop.source=source [03:50:44] item=Item(site, 'Q22') [03:50:49] item.claim(prop) [03:51:02] and that would add the property plus the source [03:53:30] Hazard-SJ: ^ ? [04:00:53] * Hazard-SJ looks up [04:05:34] legoktm: If I understood your examples correctly, it seems okay to me [04:05:48] any suggestions? [04:06:06] its more pythonic than pywikipediabot-ic [04:06:08] But none of those two classes are in the proposal :S [04:06:20] oh right [04:06:24] Property = PropertyInstance [04:06:25] -ic? [04:06:29] Item = ItemPage [04:06:34] ill work on better names later [04:07:19] I need a good way to distinguish a general "property" aka Property:P1 [04:07:20] vs [04:07:32] on Q42, P1-->Q1 [04:10:51] legoktm: Where Q42 is the item, P1 is the property, and Q1 is the value of the property? [04:10:59] Yes. [04:11:16] However I'd like the change to be transparent. [04:11:21] It might need to be one class. [04:12:36] legoktm: So rewrite already has things for that? [04:12:48] No [04:12:52] I'm going to write it :P [04:13:31] :P [04:13:56] Trunk has a function to set claims IIRC [04:14:18] This is going to be better. [04:14:32] Also,! [04:14:43] DataSite in rewrite is completely broken [04:14:52] I can't believe no one bothered to fix it. [04:14:53] legoktm: Why not improve trunk while you're at it then? :P [04:14:57] I hate trunk [04:15:01] * DataPage [04:15:03] I want to kill it. [04:15:05] No, [04:15:06]