[11:20:25] I have looked at the "Waterfalls in Australia" Wikipedia entries, and see you do not have a photograph of the Agnes Falls in Victoria. How can I send you one? [11:21:16] https://commons.wikimedia.org/wiki/Category:Agnes_Falls yes we do ;-) [11:22:37] category added on the wikidata entity [11:22:45] If you have a photo of Agnes Falls can you add it to the "Waterfalls in Australia" site, then? [11:24:40] I am the Secretary of the community group "Friends of Agnes Falls", and we have many photos of these falls on our facebook page of that name. [11:24:48] done too. [11:25:18] I will have a look! [11:25:25] as for uploading pictures, you need to understand only the author of a picture has the right to upload it, unless they already placed it under a licence compatible with Wikimedia Commons [11:25:57] and if someone rightfully uploads a picture, that picture will be placed under a free licence compatible with Wikimedia Commons [11:26:28] The photo on our FB page - the small one which is the "profile pic" - is one that I have taken myself. I am the administrator of the FB page for the group. [11:26:34] here is a bit more information https://commons.wikimedia.org/wiki/Commons:First_steps/Uploading_files [11:26:54] KayCee being administrator doesn't give you rights on the intellectual property of others ;) [11:26:58] OK. I will have a look. [11:27:05] but you can upload the pictures you've taken yourself :) [11:27:31] Thankyou. How do I send it through to you? [11:28:28] you don't, you upload it yourself, everything is explained either on the page i gave a link to or on the Upload Wizard (which is linked from that page) [11:28:50] it's been designed to be reasonably user-friendly ;) [11:29:24] Thankyou very much. I will have a look at the link and see which photo of mine would be a good one to upload. [11:29:59] thanks :) [11:30:49] If I do not do it shortly I will give it a go tomorrow. Thankyou for your advice. [11:32:01] as for adding images to wikidata, even though the relevant [[Category:Agnes Falls]] is already added to the entity, you would add a statement, specify the property Commons category (it will autofill), then give the name of the category on Commons, in this case "Agnes Falls" (without the leading "Category:") [11:32:02] 10[1] 04https://www.wikidata.org/wiki/Category:Agnes_Falls [11:33:26] obviously, you won't have to do it for the photos you add, it's already taken care of - i did however specify an image on the entity in wikidata (another statement with another property), feel free to change that if you will :) [11:35:09] I will give it a try tomorrow or the next day. It is getting quite late here in Australia now, and I will need to follow your directions more clearly. Goodbye for now. [20:49:05] SMalyshev, how hard would it be to set up a Wikidata Sparql (dynamicly updated) clone that also has a separate set of data (OSM)? This way you could query "give me a list of OSM objects' whose wikidata tag is a link to a disambig page" [20:49:55] or is it possible to setup a sparql instance only with the additional data, and so that SPARQL would work cross-instances? [20:50:22] oh, and btw, don't forget to vote in the election :D [21:13:13] yurik: not very hard. But if you have federation, you can just use that. slower of course [21:13:39] SMalyshev, does wikidata sparql supports federation? [21:13:46] yurik: yes. [21:14:15] yurik: https://www.mediawiki.org/wiki/Wikidata_query_service/User_Manual#Federation [21:14:44] i wonder how efficient it would be though... there is multi-GB OSM database -- with all the tags. If we put that data into a sparql engine, it would be great to federate that info somehow... [21:14:54] if you have an endpoint you need to add, nominate it here: https://www.wikidata.org/wiki/Wikidata:SPARQL_federation_input (or just tell me :) [21:14:55] btw, does it support geo lookups? [21:15:11] yurik: doees what support geo lookups? [21:15:18] sparql engine [21:15:35] i mean - does it have an r-tree like indexes? [21:15:35] yurik: depends on the engine :) ours does [21:16:34] hmm... i wonder if it would make sense to even host it on WMF infrastructure - all of the OSM data (~40GB with all the geometries, but we can strip that) [21:17:24] each object is a subject with a mandatory "type" property (node/way/relation), plus all the tags as properties (name of the property being an ID) [21:17:56] i'm not exactly sure how properties like "name:en" should be stored [21:18:17] doesn't wmf already have a copy of osm data to generate maps.wikimedia.org? [21:18:19] yurik: hmm not sure I can say without looking at what is in the db now. it's not in RDF now is it? [21:18:48] nikki, it does, but its not used in any sparql endpoints [21:19:10] SMalyshev, no, its a giant protobuf download, that gets parsed out into postgress [21:19:25] for this usecase, we wouldn't want to deal with the postgres, as it is mostly for geometries [21:20:13] oh, i just found http://linkedgeodata.org/About [21:20:42] SMalyshev, apparently they have a sparql endpoint! can we federate them? [21:21:32] the interface is not very nice :( [21:22:00] yurik: if they have suitable license, we sure can [21:23:43] Entities and tags undergo a filtering procedure (explained here), therefore not all available OSM data is being loaded into the store. [21:23:49] I hope it's ok [21:24:07] yurik: anyway, add it to https://www.wikidata.org/wiki/Wikidata:SPARQL_federation_input and I'll add the endpoint [21:24:49] SMalyshev, does it mean that i will be able to write a query at query.wikimedia.org, and use the data from osm? [21:24:52] or the other way around? [21:25:02] wehave much better UI :) [21:25:19] yurik: generally both. the other way around you should be able to do right now [21:25:47] to add it to query.wikidata.org I need to add endpoint to whitelist [21:36:51] SMalyshev, seems like their system has been up for a while, and is well maintained... Looking for endpoint... [21:36:57] http://linkedgeodata.org/OSM -- some query examples [21:37:14] could it be http://linkedgeodata.org/sparql ? [21:40:50] yurik: yeah looks like it [21:45:15] SMalyshev, i just added them to the https://www.wikidata.org/wiki/Wikidata:SPARQL_federation_input [21:53:14] yurik: cool, thanks [23:17:42] SMalyshev, i keep thinking if their ontology is too complex... Is it possible to use arbitrary strings as property IDs in SPARQL? [23:22:34] yurik: what do you mean? [23:22:44] not sure I understand [23:23:34] SMalyshev, given an arbitrary JS object, e.g. { "name:en":"blah", "height": "10" }, can it be stored and queried by sparql as is? [23:24:32] or should we create some magical structures like [ { "key": "name:en", "value": "blah"}, ...] [23:25:04] The number of unique keys in OSM is about 65k [23:25:21] because everyone keep creating arbitrary keys [23:53:22] yurik: well, sparql doesn't work this way [23:53:39] it's RDF. meaning you have triples - subject-predicate-object [23:53:45] i know that :) [23:53:55] i meant this object to belong to one abstract entity [23:54:02] so you'd have something like osm:nameEn "blah" [23:54:09] right [23:54:19] I have no idea though how OSM ontology looks like - they probably have some docs on it [23:54:23] but can osm:(X) be anything? [23:54:34] i mean - can X be any string, including spaces? [23:54:35] ah, yes, surely [23:54:40] no, no spaces [23:54:49] it's IRI which is an extension of URI [23:55:02] basically, URL is fine, don't go beyond that [23:55:02] so it has to be URI encoded then [23:55:12] ok, that could work [23:55:23] e.g. OSM:Blah%20something [23:55:45] or query encoded is better? Blah+something [23:55:47] hmm not sure about encoded, probably yes but may be problematic for other formats... I wouldn't use anything funky there [23:56:14] SMalyshev, its not up to me :) The whole idea is to copy the entire OSM as is, and sadly people use all sorts of strange chars in the key [23:56:40] the vast majority of them are chars and ":" symbol [23:56:53] but others also exist [23:56:58] well, https://www.ietf.org/rfc/rfc3987.txt defines it [23:57:38] but that's just internal. not sure about how formats see it [23:57:48] if it defines the valid chars, does it also define escaping? [23:58:17] seems like that's not part of it [23:58:23] is there anything stopping you from using literals instead of IRIs in a predicate? e. g. osm:whatever "foo"^^osm:blob osm:bar [23:58:42] ?? [23:58:45] just technically, I mean (I’m sure it’s a terrible idea) [23:58:48] yeah looks like IRIs accept percent encoding [23:59:23] ah, ok, that would work. As long as i can encode all these: https://taginfo.openstreetmap.org/keys [23:59:57] if you go to any random page after the first 50, you may see all sorts of strange chars there :)