Re: [OSM-talk] Can wikidata links help fight name inflation?
On Sun, Jun 7, 2015 at 7:43 PM, Simon Poole si...@poole.ch wrote: tl;dr version: linking to wikidata is probably ok, including wikidata could be a minefield. I don't think anybody was actually suggesting to include bits and pieces of Wikidata into *the* OSM database. I think the idea is for third parties to use Wikidata as a complementary source to fill in the bits and pieces that are not in OSM (either by design or by incompleteness). Of course, those third parties are well advised of the legal minefield that you mention. On the other hand, I don't think OSMF also warrants that the OSM database is really fully ODbL-compliant in its entirety all the time. If some copyright violations were added and then discovered, the DWG will revert it and further redact it if necessary. But somebody may have gotten the tainted minutely changesets before the violation is discovered. So the advice of being wary of Wikidata's CC0 license should also be the same advice for OSM's ODbL license. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Two remarks on the the discussion: - the standard point: adding translations of names to OSM is (naturally) nonsense, adding names commonly in use in a language for places isn't. I somehow suspect that Frederiks suggestion is actually an attempt to offload the dealing with the nonsense aspect to wikidata. - while superficially the licence of wikidata is claimed to be CC0, the WMF does not actually warrant (warrant as in they would cover any costs and damages if there was trouble) that this is the case. Which is naturally a concern for OSM for content derived from third party sources (typically google). Matter of fact it is a bit of a circular argument because the licensing of CC0 is based on the WMF legal marketing statement that they believe facts are not copyrightable. However we know this is not necessarily correct (using copyright in a loose sense for any similar rights) for collections of facts. tl;dr version: linking to wikidata is probably ok, including wikidata could be a minefield. Simon signature.asc Description: OpenPGP digital signature ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
2015-06-07 14:43 GMT+03:00 Simon Poole si...@poole.ch: Two remarks on the the discussion: - the standard point: adding translations of names to OSM is (naturally) nonsense, adding names commonly in use in a language for places isn't. I somehow suspect that Frederiks suggestion is actually an attempt to offload the dealing with the nonsense aspect to wikidata. This basically isolates the problem with understanding. Somehow you distinguish names from translations. There is an object. People in culture A call it A1, some of them call it A2. People from culture B call it B1. People from culture C don't use this object very often, but call it C1. People from culture D haven't ever seen this object yet, but after first one of them sees it, he calls it D2, but later scientists from culture D ask everyone to call it D1. Object is located in a way so culture B lives around it. So, in my perfect OSM world: name = B1 name:B = B1 name:A = A1 alt_name:A = A2 name:C = C1 name:D = D1 alt_name:D = D2 note: name:B is there even though culture B surrounds it - think of McDonalds in Belarus. It has name=McDonalds, even though usually objects in BY have name= in Russian. http://en.wikipedia.org/wiki/Translation says Translation is the communication of the meaning of a source-language text by means of an equivalent target-language text. Translation is about giving some valid C1 given D2. The process of matching is translation, not C1. When I see text in Russian reffering Лондон, when I retell it to someone English-speaking, I use London. (then London is translation and should be removed from OSM as nonsense ;) For geographical names there are different means that include transcription, transliteration and adaptation of all kinds. The fact that it can be thought to be automatizable to a certain degree should not trick you into thinking oh, these aren't real names. -- Darafei Komяpa Praliaskouski OSM BY Team - http://openstreetmap.by/ xmpp:m...@komzpa.net mailto:m...@komzpa.net ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Am 07.06.2015 um 14:12 schrieb Eugene Alvin Villar: .. So the advice of being wary of Wikidata's CC0 license should also be the same advice for OSM's ODbL license. The difference is that while we don't warrant that the OSM dataset is completely free of incompatible data, it is the intent and such data will be removed if identified. This isn't a panacea, but given the US DCMA and similar laws in other countries, probably good enough. There is AFAIK no such policy by the WMF, and as I said their (marketing) policy is that the problem can't exist. SImon signature.asc Description: OpenPGP digital signature ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On Sun, Jun 7, 2015 at 5:12 AM, Eugene Alvin Villar sea...@gmail.com wrote: On Sun, Jun 7, 2015 at 7:43 PM, Simon Poole si...@poole.ch wrote: tl;dr version: linking to wikidata is probably ok, including wikidata could be a minefield. I don't think anybody was actually suggesting to include bits and pieces of Wikidata into *the* OSM database. I think the idea is for third parties to use Wikidata as a complementary source to fill in the bits and pieces that are not in OSM (either by design or by incompleteness). Of course, those third parties are well advised of the legal minefield that you mention. I suggested that. I feel that for name translations *mirroring* wikidata *into* OSM has compelling advantages. It makes it easy for data consumers and ties the data where it belongs. It's pretty easy to mechanically maintain. Should spam data arrive via Wikidata/Wikpedia, it can leave via the same mechanism. If OSM is to be a language neutral dataset, no given language should be special (beyond the keys and values which are English by convention, but need not be shown to mappers or readers that way). The legal minefield can be resolved, especially with OSM and Wikmedia Foundation as organizations with a very compatible mission. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On Sun, Jun 7, 2015 at 10:29 PM, Simon Poole si...@poole.ch wrote: Am 07.06.2015 um 14:12 schrieb Eugene Alvin Villar: .. So the advice of being wary of Wikidata's CC0 license should also be the same advice for OSM's ODbL license. The difference is that while we don't warrant that the OSM dataset is completely free of incompatible data, it is the intent and such data will be removed if identified. This isn't a panacea, but given the US DCMA and similar laws in other countries, probably good enough. There is AFAIK no such policy by the WMF, and as I said their (marketing) policy is that the problem can't exist. http://wikimediafoundation.org/wiki/DMCA_Policy ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
I was the person who adds most of the labels to London, so I want to share also my thoughts. I did this for all capitals and countries in the world to be able to create worldmaps in different languages. Such world maps in local languages are known from each school atlas and are used in local television news. For this action I use Wikipedia article names. So London was in each language important enough to write an article about it in this language. Which is a lot of work, someone invested. Before this action I checked and discuss that there was no legal issue by using Wikipedia article names. Now they are at Wikidata under CC-0 so definitely free. The bias in this discussion that English labels are ok and helpful but other languages are not, is scary for me. Please take a look at: https://en.wikipedia.org/wiki/List_of_languages_by_number_of_native_speakers and find that English is only on place 3 of the list. We have a good chance that we will see in short-therm multilingual maps in lots of websites by using vector tiles. And in this moment we really want to make it harder to add such translations? Now to Frederiks proposal: To use Wikidata is really an interesting idea to reduce duplicated work, also for more than place names. With some pre-condition we could talk more positive about Frederiks proposal: *There should be in the future an additional, official version of planet-dumps and diffs that are enrich with content from Wikidata. Such a service must be stable as the other planet dumps. This service would kill the argument that it is harder to use external database Wikidata for apps. If not we create duplicated work for developers that we want to avoid elsewhere. *We would need much more Wikidata tags in OSM, now we have only 35.000 and for wikipedia-Tags we have also only 450.000. But I have a database of 3.5 Mio Wikipedia-objects that have a coordinate. So there is work for years. *We should also discuss with the Wikidata community the scope of there project. Perhaps we can come to a point where everything is inside the scope that is interesting for a Wikimedia project AND for OpenStreetMap. But if we come with the idea that we want to store opening hours for little shops in Wikidata I'm sure that the Wikidata community would not be amused to it. (I already try it.) So before we have something of the pre-conditions on the horizont I think it's easier go further with the way of name tags we have now for years. Tim alias Kolossos ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Phil phil at trigpoint.me.uk writes: Transliteration is something that can be done at application level, and a traveller would have learned the basics. You still need to be able to check the street names against.what is on the sign. Then why not have a single transliteration? A single cryllic to latin transliteration will serve all languages using the latin alphabet, do we need separate Russian, Ukrainian, Serbo Croat tags when they are identical? Transliteration is from one script to another, but many languages may share the same script, each with its own alphabet. Thus there are almost always multiple valid transliterations. It gets worse, especially when transliterating to or from non-European scripts: https://www.openstreetmap.org/user/Minh%20Nguyen/diary/35163 If you need to wayfind with OSM data, the unqualified `name` tag is your friend: as the primary name of a feature, it stands the greatest chance of appearing on signage. In any case, we shouldn't limit our imagination to travelers carrying OSM field maps or GPS units while roaming the English countryside. That use case certainly calls for consistency with signposts, but not at the expense of other use cases, such as OSM-based world maps, geography textbooks, board games, etc. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Am 31.05.2015 um 23:19 schrieb moltonel 3x Combo molto...@gmail.com: In both cases, it doesn't look like a take that routers and renderers would want to take into account. why not? both seem to cover exactly the topic of on the ground signage that we discuss in this deviation of the thread cheers Martin ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 01/06/2015, Martin Koppenhoefer dieterdre...@gmail.com wrote: Am 31.05.2015 um 23:19 schrieb moltonel 3x Combo molto...@gmail.com: In both cases, it doesn't look like a take that routers and renderers would want to take into account. s/take/tag/ why not? both seem to cover exactly the topic of on the ground signage that we discuss in this deviation of the thread Both unsigned_ref and visible_name have obviously been chosen specifically to not be picked up by the usual renderers and routers, so these consumers should probably respect the maper's intent. In contrast to the various signed=* tags which say this is an actual name/ref, but you shouldn't display it to the end user that are intended to be used by consumers. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
SomeoneElse writes: On 29/05/2015 07:07, Andrew Hain wrote: Thank you Dave. As a British mapper I am ashamed that some people want to make the map of my country less useful, and not only to Russian speakers a long way away. Hang on, where is _anybody_ saying that? The whole point of the thread is about whether it would be possible to use wikidata to make the map _more_ useful to people who don't speak a particular language, not less. As I said yesterday: It's a perfectly reasonable request for someone to ask can I have a map that shows place names displayed in my language / alphabet. It wouldn't be the first time that we argued about what the map should have on it. IMHO, we shouldn't have a map. We should have a list of map peers, each of which is rendered topically, including an English Language map, or a Kannada map. They should be hosted by whomever wants to host them. The only difference is which tile set gets selected. You can edit off all of them, using potlatch or id or remote control. There is a data flag for all of them so you can see data items rendered over the map tiles. And (this is key) there is no default. There is no OSM map. And there is nothing to argue about how the map should be rendered. We'll still argue, of course. But it will be arguments about what goes into the database. And the default answer there is Everything that doesn't move, and a few things that do. -- --my blog is athttp://blog.russnelson.com Crynwr supports open source software 521 Pleasant Valley Rd. | +1 315-600-8815 Potsdam, NY 13676-3213 | Sheepdog ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 30/05/2015, Martin Koppenhoefer dieterdre...@gmail.com wrote: there are ~600 visible_name On Overpass there's actually only two places where visible_name has been used: * In Venice, associated with a corresponding name=*, looks completely superfluous. * Somewhere else in Italy, associated with noname=yes and a note on why the signposted name isn't a proper one. In both cases, it doesn't look like a take that routers and renderers would want to take into account. 37000 unsigned_ref And 3 of those are associated with a ref=*. Not sure how this is supposed to be used, it seems like a poorly-designed tag to me. Is that inherited from TIGER ? 518 unsigned 400 signed 161 name:signed 124 name:sign 86 ref:signed 48 unsigned_name 47 ref:unsigned 16 name:signposted Yup, founds those too after Simon explained things a bit. It seems that for names nobody cares to say whether they are signposted or not (or maybe only when the sign is different from the actual name), while for ref it seems common practice(?) to use unsigned_ref I wouldn't trust or interpret unsigned_ref until I understand a bit more about it. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 2015-05-31 14:19, moltonel 3x Combo wrote: On 30/05/2015, Martin Koppenhoefer dieterdre...@gmail.com wrote: It seems that for names nobody cares to say whether they are signposted or not (or maybe only when the sign is different from the actual name), while for ref it seems common practice(?) to use unsigned_ref I wouldn't trust or interpret unsigned_ref until I understand a bit more about it. `unsigned_ref` is used almost exclusively in the U.S. [1] for certain kinds of route numbers that are of relatively little use to the general public. For example, the state legislature or highway department may have assigned a road or bridge a number for accounting purposes. In many cases, the numbers are signposted, but only in fine print on signage or on small signs out of view to most motorists. This tag was most recently discussed on talk-us back in January regarding Pennsylvania's quadrant route system. [2] Here's a real world example of `unsigned_ref`: https://www.flickr.com/photos/dougtone/4176740362/ What you see in this photo is typical signage for Pennsylvania State Route 443 (ref=PA 443); below it is mile marker 210. At the top of that mile marker, in infinitesimal print, is the quadrant route designation SR 3009 (thus unsigned_ref=SR 3009). Most likely, the bridge itself has a reference number carved into either end (unsigned_ref again, I guess). Some other states and counties handle things very differently, avoiding signposting legislative route or bridge numbers at all. But I'd be surprised if someone has actually bothered to import such numbers in large quantity. It was important to avoid putting these numbers in `ref` because renderers and routers alike would immediately present them as ordinary route numbers and probably give laypeople a lot of confusion. (It'd be as if someone had set `name=Londres` for London.) Other than that, the choice of unsigned was just a matter of convenience and plainly a misnomer. Something like `ref:penndot` may've been more appropriate in examples like the one above. [1] http://taginfo.openstreetmap.org/keys/unsigned_ref#map [2] https://lists.openstreetmap.org/pipermail/talk-us/2015-January/014157.html -- m...@nguyen.cincinnati.oh.us ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
We've got beautiful name:LANGUAGE tags, they work, I say let's use them. The English speaking community is well started into this practice (just look at the 1.3 million name:en tags [1]!) - now let's have the rest of us get in on the fun too ;-) Frederik's called upon the local-first rule a couple of times. It's a really useful principle for deciding conflicting edits, and I read it as local observation wins. This is good, but it shouldn't mean if you can't read the name of a thing at its place in a specific language it doesn't have one - just like so many other things that aren't labeled in the real world in the local language. In regards to relying on Wikidata for translations: OpenStreetMap is so much bigger than the people who pay attention to the mailing list. If we changed the standing practice of name:LANGUAGE tags and punted translations to Wikidata - how would the average mapper know about it? Would we write Wikidata integration for iD and JOSM to allow for seamless editing of Wikidata translations, then migrate name:LANGUAGE tags off to Wikidata? This seems a lot of effort for unclear gain. [1] Others have already compared the usage of name:en vs name:ru http://taginfo.openstreetmap.org/search?q=name%3Aen http://taginfo.openstreetmap.org/search?q=name%3Aru On Fri, May 29, 2015 at 8:27 AM, SomeoneElse li...@atownsend.org.uk wrote: On 29/05/2015 12:58, moltonel 3x Combo wrote: On 29/05/2015, SomeoneElse li...@atownsend.org.uk wrote: I'd say that whether or not a name is actually usable to help navigate to a place is a pretty important piece of information. True. And that what name should give you. The name:CC tags should not be a hindrance, at most they should be given alongside name by your satnav. For example, when processing OSM data for my own use I'll try and drop unsigned names and refs from roads (there's no point in saying turn left on Foo Street if Foo Street does not appear on the sign). That's really neat. How do you know wether a street is signposted or not ? I don't know of any tag that gives that info. I can't imagine a good heuristic using name:CC. I've added quite a few unsignposted street names by asking locals. I've used name:signed=no (though this is by no means an accepted tag, and if anyone can come up with a more accepted version that does the same job I'm all ears). Maybe something like name:signed=en;cy might solve the name verifiability problem for Abergavenny? It's the same principle here - if there are 300 names for a place, are you really suggesting that I have to do an external check to some other database to find that as it's in South Wales, signs are likely to be in Welsh and English, so it's those language names that I need to look out for? What's wrong with name ? What's the UK policy on the content of name for places with Welsh and English names ? If you want to see Welsh names as often as possible but still make the local name more prominent, use local name (welsh name if different) in your map generating script. The problem here is that there are two* equally valid and correct names (in on-the-ground verifiable terms) for Abergavenny. Obviously there are more complicated places too - http://www.openstreetmap.org/node/52241235 and http://www.openstreetmap.org/node/267762522 immediately sprung to mind (re the former I thought that it was An Daingean that appeared on the official signs these days?). Cheers, Andu * or possibly three if you count Latin. It wouldn't surprise me if that wasn't on a Welcome to Abergavenny sign somewhere. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
I find it odd that inaction, incapability or incompetence of local sign installers is a worry for a database of geographical facts, which OSM is. Some roads are hundreds of miles long, at what interval does it need to be signed for it to be considered signed. Beginning and end? Every 100 km? Some streets have signs that are now obscured by trees and bushes. On the ground rule is a rule of thumb, not the entirety of factual truth. People similarly being lax in making sure their homes conform to local rules about address visibility (huge pet peeve of mine since being a paper boy in the 1980s) should not hinder us in giving their homes the proper address in our database of facts. If you wish to use on the ground as the only rule then we can just forget about the whole thing - defeated by inaction of people that should be putting up signs and maintaining them. As for the name inflation, there is no such thing. Illegal imports maybe but name inflation as a problem does not exist. Any talk of too much data by adding an extra name field to a node is defeated soundly by any 3d mapping entity. I've looked at some of the 3D stuff done and it blows my mind how detailed and cool it is, then when I peek at the code behind it I admit defeat in trying to understand it at a glance as it is an intricate series of relations upon relations. A simpler example is Stade de France, fantastic stadium - was there myself at 1998 World Cup. Data bloat can happen, Wikidata is too fragile, we need our own store of data (POI extra details, names, translations etc) that can exchange material with Wikidata but is under full OSMF control - and not controlled by notability deletionists. In 5 years time would we be arguing about bloat in OSMData as someone suddenly purges some section from it? /rambling --Jói / Stalfur Þann 30.5.2015 12:22, skrifaði Martin Koppenhoefer: Am 29.05.2015 um 13:58 schrieb moltonel 3x Combo molto...@gmail.com: That's really neat. How do you know wether a street is signposted or not ? I don't know of any tag that gives that info. there are ~600 visible_name 37000 unsigned_ref 518 unsigned 400 signed 161 name:signed 124 name:sign 86 ref:signed 48 unsigned_name 47 ref:unsigned 16 name:signposted I haven't checked on which kind of objects or which do not refer to highways or names Very likely I also missed some variants It seems that for names nobody cares to say whether they are signposted or not (or maybe only when the sign is different from the actual name), while for ref it seems common practice(?) to use unsigned_ref ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Am 29.05.2015 um 14:05 schrieb SomeoneElse li...@atownsend.org.uk: * I have absolutely no idea what signage is around Haifa - if a name:en for a place does appear on signs and is useful to use to see if you have got there then clearly there needs to be some indicator that says that name:en is useful here, but (say) name:cy is not. why are we focussing only on signs? What kind of signs? There are hundreds of name:en tags in osm for features in the municipality of Rome, but you won't find an official sign that says Rome, you might find quite a lot of touristic information boards with English names at the monuments, usually just English and Italian, but in OSM we do have typically also de, fr, es, ru and some more localized names tagged. All of them are helpful if you speak these languages, naturally, and they are in use - not on signs but in travel guides (books), used by tour guides (humans), tourists etc. If you speak to English folk they will speak about Tuscany rather than Toscana. On the ground truth for me is not just signs, it is also oral knowledge, people talking to each other, stuff that is in books or newspapers talking about the ground, etc Cheers Martin cheers Martin ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Am 29.05.2015 um 13:58 schrieb moltonel 3x Combo molto...@gmail.com: That's really neat. How do you know wether a street is signposted or not ? I don't know of any tag that gives that info. there are ~600 visible_name 37000 unsigned_ref 518 unsigned 400 signed 161 name:signed 124 name:sign 86 ref:signed 48 unsigned_name 47 ref:unsigned 16 name:signposted I haven't checked on which kind of objects or which do not refer to highways or names Very likely I also missed some variants It seems that for names nobody cares to say whether they are signposted or not (or maybe only when the sign is different from the actual name), while for ref it seems common practice(?) to use unsigned_ref cheers Martin ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Am 29.05.2015 um 15:55 schrieb Komяpa m...@komzpa.net: What exactly are we trying to save by omitting these locales, what wasn't eaten already by source= on French buildings? :) yes, and there are more tags that are both, used more often and are more questionable than name:ru tags, e.g. tiger:name_base tiger:name_type osak:street_name osak:municipality_name addr:street:name even of tiger:name_base_1 there are more than a million, almost double the amount of all Russian names in osm. These are followed by a lot crap in the same quantity order, like KSJ2:filename (!) ~500K All these came to light by a single taginfo search for name, needless to say similar import crap can be found for different keys as well. Many of them are not even documented. cheers Martin ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Hi, Komяpa: What exactly are we trying to save by omitting these locales, what wasn't eaten already by source= on French buildings? :) Martin Koppenhoefer: even of tiger:name_base_1 there are more than a million, almost double the amount of all Russian names in osm. Firstly: It is a common fallacy for people in OSM to argue: Someone else is doing stupid things, and while they do that, everyone else should certainly be allowed to do stupid things also. I don't buy into that logic; just because silly (or sillier) stuff happens elsewhere in OSM doesn't mean that lesser problems should automatically be ignored. Secondly: I am less concerned about rubbish remaining from imports because that can be cleaned up without someone complaining. I am more concerned about things that become established and all of a sudden you find there's no way back because people rely on it. KSJ2:filename is not one of these tags. Thirdly: My main problem with name tag inflation is not the data volume (even though I can see this becoming an issue if someone should really argue that each named object has a name in each of the several thousand langauges on the planet!). My problem is that people start labelling places they have zero knowledge of, relying on military atlases from 50 years ago, or dubious travel websites, or transliterated guesswork, or database lookups. In my opinion this is often not verifiable, at least not for the usual sense in which we use verifiable. Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
I don’t think that it’s a good idea to rely on a external database. (inconsistency, overhead for map producers/developers, different toolchains, etc.) If the name:xx is just a simple transliteration you won’t need Wikidata, you can use algorithms for producing a map in the foreign language. If the foreign name differs from the transliteration I think no one would oppose to add the name into the OSM database. -- It isn’t necessary to add Wikidata tags for naming purposes. Am 28.05.2015 um 23:27 schrieb moltonel 3x Combo: Please handle the questions of should FOO-language name of an object be allowed in the database ? and should that databse be OSM or Wikidata ? separately. The decision of whether Абергавенни should be recorded as the Russian name for Abergavenny should be the same for Wikidata and OSM. +1 ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
2015-05-28 23:00 GMT+02:00 Andy Mabbett a...@pigsonthewing.org.uk: On 28 May 2015 at 09:50, Martin Koppenhoefer dieterdre...@gmail.com wrote: e.g. the en:Spanish Steps / de:Spanische Treppe are called Scalinata di Trinità dei Monti in the local language (it is located at piazza di Spagna, that's where the foreign name comes from, while in Italian it is called after to church it leads to). Naturally, OSM has the original name of this world famous monument, but Wikidata hasn't. It does now. OK, this is one point for you, but it also proves my point: wikidata at the moment is not sufficiently mature (IMHO) to replace name tags in different languages in OSM. Of course you can fix wikidata issues (if you understand how it is done, I haven't had enough time yet to understand how to make edits like this, and the fact that not all tags are shown to me (e.g. only 4 out of all language labels, after I have explicitly clicked on in more languages)) doesn't help. // sidenote: Now I could link the wikidata object of the spanish steps to the OSM object and get an Italian name. But I will not have an Italian wikipedia article about it, because it is covered in the spanish square (piazza di spagna) article in Italian. How would I ideally procede now? a) in wikidata link the article of the piazza di spagna (in italian) to the wikidata object about the spanish steps? a2) like a) but link to an anchor: Piazza_di_Spagna#La_scalinata ? b) in wikipedia split the italian wikipedia article in 2, one for the square and one for the steps? c) in osm add an additional tag like wikipedia:it=Piazza_di_Spagna#La_scalinata to the steps object? d) something different... //sidenote off If we were to massively use wikidata _instead of duplicating some details from there also in our db_ we would have to improve wikidata as well, You'd be welcome to do so. ... and impose our entity structure on them, Really? Good luck with that. what I meant, and what you do confirm below: if for instance there is an object in wikidata which is an administrative entity and a geographic place at the same time, but for OSM we'd need 2 distinct objects, we will have to split the wikidata object. This could be done only if there wasn't resistance from other wikidata users who might want to keep the current unmodified object because it links better to wikipedia articles. We might introduce another object that linked the split objects onto one, which could serve for wikipedia articles, etc. but this is a much more complicated procedure than changing tags in OSM alone. We've always said that we wanted editing to be simple, so that we can maximize the amount of available editors, but with the tight integration of another dynamic dataset (for one of the core competences we are dealing with: toponyms) or it won't work in some cases (and if it doesn't work in some case, it doesn't work at all). That is, of course, nonsense. OK, let's say it is nonesense, because you can accept that a solution works for most of the cases and try work around those that don't work. Currently (all names in OSM) we don't have these problems though. Another issue I see with wikidata is that it contains information and details about spatial objects, but it doesn't contain the geometry it refers to. The geometry is in OSM, is it not? Why would Wikidata want to replicate that? IMHO you have to understand to which geometry you are referring when you make edits, or you might break stuff without noticing it. Wikidata editors would have to look at OSM geometries to ensure that their edit maintains consistency, and OSM users would have to check wikidata to see if editing something in a certain way (e.g. merges or splits, adding tags, changing geometry) is OK or whether they have to split the wikidata object and update the wikidata link. It is not impossible, but it is an enormous amount of complexity added, and it also augments the risk of non-availability of the backend by 100% (because now we depend on 2 services and not on one). I want to point out is that there seem to be different criteria defined for different languages: These descriptions aid users; they are not proscriptive. There are also local and cultural variations. Just like city in OSM. Maybe these are descriptions to aid in some regional wiki projects and proscriptive rules in others like Germany, where rules rule? Just like in OSM ;-) It would rather confuse than aid me if the descriptions in some language says something is foo and in another language they tell me it is not foo. TL;DR; wikidata is a gorgeous project, combining their knowledge with ours is very promising. Still, in my opinion, for the current state of where they are (and where the tools to combine both are), I would _not remove tags_ from OSM just because the same information might be available in wikidata. Cheers, Martin ___ talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 29/05/2015 07:07, Andrew Hain wrote: Thank you Dave. As a British mapper I am ashamed that some people want to make the map of my country less useful, and not only to Russian speakers a long way away. Hang on, where is _anybody_ saying that? The whole point of the thread is about whether it would be possible to use wikidata to make the map _more_ useful to people who don't speak a particular language, not less. As I said yesterday: It's a perfectly reasonable request for someone to ask can I have a map that shows place names displayed in my language / alphabet. Cheers, Andy ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Dave Corley davecorley at gmail.com writes: Lastly, and I think this is important point. To quote the wiki header . the project that creates and distributes free geographic data for the world. Either this is a database of worldwide geodata or its not. There's no half-way in that statement. Either all cultures, languages, countries, people and the variety these elements bring in terms of tagging, is accepted on a universal basis or its not. Thank you Dave. As a British mapper I am ashamed that some people want to make the map of my country less useful, and not only to Russian speakers a long way away. -- Andrew ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 29/05/2015 12:51, Maarten Deen wrote: It depends on what you want. When someone asks me to navigate to Natanzon, Haifa how can I enter it when the only name in the map is נתנזון, חיפה? I don't see why the transliterated name is not important. No-one is saying that the transliterated name is not useful for all sorts of reasons (the thread title can wikidata links help... makes it clear that this is about trying to make it easier to get to these names, not harder). The question here is, in a case when a name:xx isn't widely used in the place and doesn't appear on signs*, how can a user of the data know that they've got there or not? OSM shouldn't be it's own parallel universe - we map what's on the ground. Cheers, Andy * I have absolutely no idea what signage is around Haifa - if a name:en for a place does appear on signs and is useful to use to see if you have got there then clearly there needs to be some indicator that says that name:en is useful here, but (say) name:cy is not. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 29/05/2015, Maarten Deen md...@xs4all.nl wrote: It depends on what you want. When someone asks me to navigate to Natanzon, Haifa how can I enter it when the only name in the map is נתנזון, חיפה? Your software should know about both names (so that you can search for either) and show you the local name (so that you can compare to streetsigns) in addition to the name in a language of your choosing (so that you know what you're doing). And yes, I see the combination of name and a long list of name:CC as a good backend for that requirement. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 29/05/2015 12:14, moltonel 3x Combo wrote: On 29/05/2015, SomeoneElse li...@atownsend.org.uk wrote: how do we distinguish in the Abergavenny case between the two established names and the up to 7,000 (but realistically in the short term a few hundred) translations? That's unfortunately something that name:xx in OSM doesn't give us currently You don't distinguish them, and that's fine. There may be many many more pleople using Abergavenny than Абергавенни, but that doesn't mean that the name isn't established, at its more limited scale. I'd say that whether or not a name is actually usable to help navigate to a place is a pretty important piece of information. For example, when processing OSM data for my own use I'll try and drop unsigned names and refs from roads (there's no point in saying turn left on Foo Street if Foo Street does not appear on the sign). It's the same principle here - if there are 300 names for a place, are you really suggesting that I have to do an external check to some other database to find that as it's in South Wales, signs are likely to be in Welsh and English, so it's those language names that I need to look out for? Cheers, Andy ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 29/05/2015, SomeoneElse li...@atownsend.org.uk wrote: I'd say that whether or not a name is actually usable to help navigate to a place is a pretty important piece of information. True. And that what name should give you. The name:CC tags should not be a hindrance, at most they should be given alongside name by your satnav. For example, when processing OSM data for my own use I'll try and drop unsigned names and refs from roads (there's no point in saying turn left on Foo Street if Foo Street does not appear on the sign). That's really neat. How do you know wether a street is signposted or not ? I don't know of any tag that gives that info. I can't imagine a good heuristic using name:CC. I've added quite a few unsignposted street names by asking locals. It's the same principle here - if there are 300 names for a place, are you really suggesting that I have to do an external check to some other database to find that as it's in South Wales, signs are likely to be in Welsh and English, so it's those language names that I need to look out for? What's wrong with name ? What's the UK policy on the content of name for places with Welsh and English names ? If you want to see Welsh names as often as possible but still make the local name more prominent, use local name (welsh name if different) in your map generating script. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 29/05/2015, p...@trigpoint.me.uk p...@trigpoint.me.uk wrote: On Fri May 29 11:27:41 2015 GMT+0100, moltonel 3x Combo wrote: Even if it was justified, the reversals were either sloppy or biased : see http://www.openstreetmap.org/node/10021976/history which got name:ru Лестер deleted but was allowed to keep name:ukЛестер. The Ukrainian names were added after the mapper engaged the UK community. If that's the case, it means the reversal was biased rather than sloppy, which is arguably worse. name:ru was first added 4 years ago, name:uk 1 year ago. The more recent changesets that re-added name:ru got a changeset discussion where the contributor explained his verification process, which isn't much but is probably as good as can be in this case. Лестер is the same message from Ukraine and from Russia, but in the second case the messenger was shot. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 2015-05-29 13:34, SomeoneElse wrote: On 29/05/2015 12:14, moltonel 3x Combo wrote: On 29/05/2015, SomeoneElse li...@atownsend.org.uk wrote: how do we distinguish in the Abergavenny case between the two established names and the up to 7,000 (but realistically in the short term a few hundred) translations? That's unfortunately something that name:xx in OSM doesn't give us currently You don't distinguish them, and that's fine. There may be many many more pleople using Abergavenny than Абергавенни, but that doesn't mean that the name isn't established, at its more limited scale. I'd say that whether or not a name is actually usable to help navigate to a place is a pretty important piece of information. For example, when processing OSM data for my own use I'll try and drop unsigned names and refs from roads (there's no point in saying turn left on Foo Street if Foo Street does not appear on the sign). It depends on what you want. When someone asks me to navigate to Natanzon, Haifa how can I enter it when the only name in the map is נתנזון, חיפה? I don't see why the transliterated name is not important. Maarten ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
I'd say that whether or not a name is actually usable to help navigate to a place is a pretty important piece of information. For example, when processing OSM data for my own use I'll try and drop unsigned names and refs from roads (there's no point in saying turn left on Foo Street if Foo Street does not appear on the sign). It depends on what you want. When someone asks me to navigate to Natanzon, Haifa how can I enter it when the only name in the map is נתנזון, חיפה? I don't see why the transliterated name is not important. while the transliterated / translated name might be useless for the actual navigation, it is important for entering the address. Seems like you are both addressing a different part of the task. I assume a Brit wants to enter Antwerp, Belgium and not Antwerpen, België. Although once they are driving in Belgium they will get navigational hints as follow the motorway towards Antwerpen (or maybe even towards Anvers when that is on a sign in France). similar for the Foo Street. When I want to drive to the Foo Street, I want to be able to enter it, whether is is signed or not. I'm probably fine that the driving instructions leave out the name when I get to my destination. regards m. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On Fri, 2015-05-29 at 13:14 +0200, moltonel 3x Combo wrote: And as it happens, Абергавенни comes from Abergavenny rather than Y Fenni, showing that some discernment was applied. Not sure I understand that statement, transliterating Y Fenni is equally valid in my view. Phil (trigpoint) ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Single handedly deleting hundreds of name:ru on nodes without an alternative in place instantly made map less useful. Upphafleg skilaboð Frá: SomeoneElse li...@atownsend.org.uk Dagsetning: 29/05/2015 07:53 (GMT+00:00) Til: talk@openstreetmap.org Efni: Re: [OSM-talk] Can wikidata links help fight name inflation? On 29/05/2015 07:07, Andrew Hain wrote: Thank you Dave. As a British mapper I am ashamed that some people want to make the map of my country less useful, and not only to Russian speakers a long way away. Hang on, where is _anybody_ saying that? The whole point of the thread is about whether it would be possible to use wikidata to make the map _more_ useful to people who don't speak a particular language, not less. As I said yesterday: It's a perfectly reasonable request for someone to ask can I have a map that shows place names displayed in my language / alphabet. Cheers, Andy ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 29 May 2015 at 09:58, Martin Koppenhoefer dieterdre...@gmail.com wrote: 2015-05-28 23:00 GMT+02:00 Andy Mabbett a...@pigsonthewing.org.uk: On 28 May 2015 at 09:50, Martin Koppenhoefer dieterdre...@gmail.com wrote: e.g. the en:Spanish Steps / de:Spanische Treppe are called Scalinata di Trinità dei Monti in the local language (it is located at piazza di Spagna, that's where the foreign name comes from, while in Italian it is called after to church it leads to). Naturally, OSM has the original name of this world famous monument, but Wikidata hasn't. It does now. OK, this is one point for you, but it also proves my point: wikidata at the moment is not sufficiently mature (IMHO) to replace name tags in different languages in OSM. It proves nothing of the kind, Does OSM still have work needing to be done? Is OSM not sufficiently mature to replace other mapping services? Of course you can fix wikidata issues (if you understand how it is done, I haven't had enough time yet to understand how to make edits like this, How long does it take a new mapper to understand how to make such edits in OSM? and the fact that not all tags are shown to me Try logging in and enabling the Labels list gadget; or alternatively https://www.wikidata.org/wiki/Q1?withJS=MediaWiki:Gadget-labelLister.js for the Q value of your choice (I did say I would provide an update about this in an earlier post) (e.g. only 4 out of all language labels, after I have explicitly clicked on in more languages)) doesn't help. Did you use the configure link alongside that? Now I could link the wikidata object of the spanish steps to the OSM object and get an Italian name. But I will not have an Italian wikipedia article about it, because it is covered in the spanish square (piazza di spagna) article in Italian. How would I ideally procede now? There is no Italian Wikipedia article about the steps. So what? and impose our entity structure on them, Really? Good luck with that. what I meant, and what you do confirm below: if for instance there is an object in wikidata which is an administrative entity and a geographic place at the same time, but for OSM we'd need 2 distinct objects, we will have to split the wikidata object. This could be done only if there wasn't resistance from other wikidata users who might want to keep the current unmodified object because it links better to wikipedia articles. There would be no such resistance. We might introduce another object that linked the split objects onto one, which could serve for wikipedia articles, etc. but this is a much more complicated procedure than changing tags in OSM alone. It's trivially easy. You just said atht you haven't yet learned to do so, that is all. or it won't work in some cases (and if it doesn't work in some case, it doesn't work at all). That is, of course, nonsense. OK, let's say it is nonesense, because you can accept that a solution works for most of the cases and try work around those that don't work. Currently (all names in OSM) we don't have these problems though. We have another set of problems: Disagreement over when to include names; insufficient volunteer-hours, etc. IMHO you have to understand to which geometry you are referring when you make edits, or you might break stuff without noticing it. Wikidata editors would have to look at OSM geometries to ensure that their edit maintains consistency, Why would they? I say that claim is bogus. and OSM users would have to check wikidata to see if editing something in a certain way (e.g. merges or splits, adding tags, changing geometry) is OK or whether they have to split the wikidata object and update the wikidata link. It is not impossible, but it is an enormous amount of complexity added, For those who wish to do so, this is not onerous. and it also augments the risk of non-availability of the backend by 100% (because now we depend on 2 services and not on one). Not if, as others have suggested, data is cached. In any case, WIkdiata's relaibality, like other WMF projets, is commendably high. -- Andy Mabbett @pigsonthewing http://pigsonthewing.org.uk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 29 May 2015 at 09:57, Arch Arch 7h3.a...@gmail.com wrote: If the foreign name differs from the transliteration I think no one would oppose to add the name into the OSM database. Then you haven't been paying attention to this thread. -- Andy Mabbett @pigsonthewing http://pigsonthewing.org.uk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
W dniu 29.05.2015 11:45, Arch Arch napisał(a): I don't think that's a good idea to try to solve the problems we're faced with by our current OSM data model by setting up a second database. We need an improved data model which fits our needs. Current OSM data model is not flexible enough for expanding for the rest of the world, and even for many not typical items in the currently mapped parts of the world. If we will still try to put categories into the tagging, we won't be able to recategorize it later for example, because there's a strong backlash against mechanical edits of multiple objects. IMO we need to let the mappers tag the ground truth and take the meta informations away, so we can process them and their relations easier (amenity or shop is so wide, and currently there's no way to make subcategories if needed). And note, that this second database (OSMdata) will be just OSM Wiki on steroids - and that is better, because quite popular opinion is that it's not manageable anymore. -- The train is always on time / The trick is to be ready to put your bags down [A. Cohen] ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 29/05/2015 10:42, Nick Whitelegg wrote: You don't even need on the ground evidence. You just need someone ... or something ... with knowledge of Cyrillic and Roman alphabets to be able to transliterate Abergavenny into the Cyrillic, presumably. Absolutely. But leaving aside whether that's created on the fly (perhaps from a stored IPA pronunciation, though initial attempts to do that haven't proved successful*) or stored (either in OSM or wikidata or elsewhere) how do we distinguish in the Abergavenny case between the two established names and the up to 7,000 (but realistically in the short term a few hundred) translations? That's unfortunately something that name:xx in OSM doesn't give us currently, though from reading Andy Mabbett's links it's something that wikidata would. A question to those suggesting Абергавенни as a name:ru for Abergavenny / Y Fenni - how to distinguish the latter (both are names that you can to compare with signposts to see that you're going in the right direction) with the former (which you can't, but a speaker of that language might use to refer to that place in their own language)? Cheers, Andy * https://metacpan.org/pod/Text::TransMetaphone::ru , which Komяpa tried last night. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On Fri, 2015-05-29 at 13:58 +0200, moltonel 3x Combo wrote: What's wrong with name ? What's the UK policy on the content of name for places with Welsh and English names ? If you want to see Welsh names as often as possible but still make the local name more prominent, use local name (welsh name if different) in your map generating script. By UK I assume you mean the UK (GB) OSM community? The UK government has no policy, such decisions are made by the Welsh Government. My interpretation of the rule as to which goes in name is whichever comes first on the signpost. The general rule is in welsh speaking areas, welsh will come first, english first in english speaking areas. Usual way to tell is if Araf or Slow comes first on the road markings. Not sure which Abergavenny is, but didn't hear much Welsh last time I was on a late night train through there. Phil (trigpoint) ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Exactly. Us Western Europeans would find Romanised versions of names useful when travelling out of Western Europe (to give a real example: I'm visiting Greece this summer, and while I'm just about at the stage where I think I can decode the Greek alphabet, Romanised versions are definitely helpful), and the converse should also be true. You don't even need on the ground evidence. You just need someone with knowledge of Cyrillic and Roman alphabets to be able to transliterate Abergavenny into the Cyrillic, presumably. Nick From: Andrew Hain andrewhain...@hotmail.co.uk Sent: 29 May 2015 07:07 To: talk@openstreetmap.org Subject: Re: [OSM-talk] Can wikidata links help fight name inflation? Dave Corley davecorley at gmail.com writes: Lastly, and I think this is important point. To quote the wiki header . the project that creates and distributes free geographic data for the world. Either this is a database of worldwide geodata or its not. There's no half-way in that statement. Either all cultures, languages, countries, people and the variety these elements bring in terms of tagging, is accepted on a universal basis or its not. Thank you Dave. As a British mapper I am ashamed that some people want to make the map of my country less useful, and not only to Russian speakers a long way away. -- Andrew ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
We need our own OSMdata instance in between to describe real world objects and concepts. And maybe we could even solve the ridiculous amount of duplication we're experiencing at the moment. True, the editor software will have to be adapted to cope with merges and splits, so the human editor can decide what OSM object(s) belong to what real world object(s). Either that or we should start using relations more intelligently. But they are heavyweight and supposedly they are also complicated. Polyglot 2015-05-29 10:58 GMT+02:00 Martin Koppenhoefer dieterdre...@gmail.com: 2015-05-28 23:00 GMT+02:00 Andy Mabbett a...@pigsonthewing.org.uk: On 28 May 2015 at 09:50, Martin Koppenhoefer dieterdre...@gmail.com wrote: e.g. the en:Spanish Steps / de:Spanische Treppe are called Scalinata di Trinità dei Monti in the local language (it is located at piazza di Spagna, that's where the foreign name comes from, while in Italian it is called after to church it leads to). Naturally, OSM has the original name of this world famous monument, but Wikidata hasn't. It does now. OK, this is one point for you, but it also proves my point: wikidata at the moment is not sufficiently mature (IMHO) to replace name tags in different languages in OSM. Of course you can fix wikidata issues (if you understand how it is done, I haven't had enough time yet to understand how to make edits like this, and the fact that not all tags are shown to me (e.g. only 4 out of all language labels, after I have explicitly clicked on in more languages)) doesn't help. // sidenote: Now I could link the wikidata object of the spanish steps to the OSM object and get an Italian name. But I will not have an Italian wikipedia article about it, because it is covered in the spanish square (piazza di spagna) article in Italian. How would I ideally procede now? a) in wikidata link the article of the piazza di spagna (in italian) to the wikidata object about the spanish steps? a2) like a) but link to an anchor: Piazza_di_Spagna#La_scalinata ? b) in wikipedia split the italian wikipedia article in 2, one for the square and one for the steps? c) in osm add an additional tag like wikipedia:it=Piazza_di_Spagna#La_scalinata to the steps object? d) something different... //sidenote off If we were to massively use wikidata _instead of duplicating some details from there also in our db_ we would have to improve wikidata as well, You'd be welcome to do so. ... and impose our entity structure on them, Really? Good luck with that. what I meant, and what you do confirm below: if for instance there is an object in wikidata which is an administrative entity and a geographic place at the same time, but for OSM we'd need 2 distinct objects, we will have to split the wikidata object. This could be done only if there wasn't resistance from other wikidata users who might want to keep the current unmodified object because it links better to wikipedia articles. We might introduce another object that linked the split objects onto one, which could serve for wikipedia articles, etc. but this is a much more complicated procedure than changing tags in OSM alone. We've always said that we wanted editing to be simple, so that we can maximize the amount of available editors, but with the tight integration of another dynamic dataset (for one of the core competences we are dealing with: toponyms) or it won't work in some cases (and if it doesn't work in some case, it doesn't work at all). That is, of course, nonsense. OK, let's say it is nonesense, because you can accept that a solution works for most of the cases and try work around those that don't work. Currently (all names in OSM) we don't have these problems though. Another issue I see with wikidata is that it contains information and details about spatial objects, but it doesn't contain the geometry it refers to. The geometry is in OSM, is it not? Why would Wikidata want to replicate that? IMHO you have to understand to which geometry you are referring when you make edits, or you might break stuff without noticing it. Wikidata editors would have to look at OSM geometries to ensure that their edit maintains consistency, and OSM users would have to check wikidata to see if editing something in a certain way (e.g. merges or splits, adding tags, changing geometry) is OK or whether they have to split the wikidata object and update the wikidata link. It is not impossible, but it is an enormous amount of complexity added, and it also augments the risk of non-availability of the backend by 100% (because now we depend on 2 services and not on one). I want to point out is that there seem to be different criteria defined for different languages: These descriptions aid users; they are not proscriptive. There are also local and cultural variations. Just like city in OSM. Maybe these are descriptions to aid in
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 29/05/2015, SomeoneElse li...@atownsend.org.uk wrote: On 29/05/2015 07:07, Andrew Hain wrote: Thank you Dave. As a British mapper I am ashamed that some people want to make the map of my country less useful, and not only to Russian speakers a long way away. Hang on, where is _anybody_ saying that? The whole point of the thread is about whether it would be possible to use wikidata to make the map _more_ useful to people who don't speak a particular language, not less. The thread opening was indeed about using wikidata translations (which would make the map more usefull for all) and perhaps deprecating name:CC in OSM as a result (which is much more debatable). But then the question came up (By Richard I think ?) of which name:CC should be kept for a particular object, and the large number of name:ru in the UK (reverted by you and others) was given as an example of an unaccepable name:CC. And then some agreed that name:en in non-English-speaking coutries should get the same treatment. This second paragraph can be seen as making the map less useful (I share that POV). It has been argued that the reverts were justified because the names were just transliterations and not real names, but I think that this was overzealous (it wasn't blind tranliteration, there was human skill involved). Even if it was justified, the reversals were either sloppy or biased : see http://www.openstreetmap.org/node/10021976/history which got name:ruЛестер deleted but was allowed to keep name:ukЛестер. And just to be clear: pushing the decision of whether Лестер is the correct russian name of Leicester to wikidata just so that OSM looks cleaner is not a solution. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 2015-05-29 12:05, SomeoneElse wrote: A question to those suggesting Абергавенни as a name:ru for Abergavenny / Y Fenni - how to distinguish the latter (both are names that you can to compare with signposts to see that you're going in the right direction) with the former (which you can't, but a speaker of that language might use to refer to that place in their own language)? IMHO the name tag should reflect what's on the ground. The Belgians do this for names in the capital Brussels: name=Bruxelles - Brussel for the name of the city or name=Rue de la Montagne - Bergstraat for a streetname. The same way, if Abergavenny is referred to on street signs as Abergavenny / Y Fenni, its name should be Abergavenny / Y Fenni. name is the official (local) name, name:ru is the name when you want to make a map in russian and can be used in Nominatim to search for a place. Maarten ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 29/05/2015, SomeoneElse li...@atownsend.org.uk wrote: how do we distinguish in the Abergavenny case between the two established names and the up to 7,000 (but realistically in the short term a few hundred) translations? That's unfortunately something that name:xx in OSM doesn't give us currently You don't distinguish them, and that's fine. There may be many many more pleople using Abergavenny than Абергавенни, but that doesn't mean that the name isn't established, at its more limited scale. Speaking of plain dumb transliterations that got established, have a read of https://en.wikipedia.org/wiki/Place_names_in_Ireland . Despite sharing the latin alphabet, all Irish placenames have been systematically translated when Ireland was under British rule. The transliteration job has been ridiculed by the locals and is no better than what a modern algorythm would do, but for better or worse today I live in Kilkenny more than I live in Cill Cainnigh. A question to those suggesting Абергавенни as a name:ru for Abergavenny / Y Fenni - how to distinguish the latter (both are names that you can to compare with signposts to see that you're going in the right direction) with the former (which you can't, but a speaker of that language might use to refer to that place in their own language)? Where the signpost toward Pékin (the french name for Beijing) ? Where on UK soil can I find the city sign for Londres ? Foreign names for local places *are* harder to verify, and we rightfully cast a more critical eye on them. But we've got plenty of hard-to-verify data in OSM, and we rarely take a deletionist approach to it. And for better or worse, Russia-based contributors are better suited to know what should go into name:ru than UK-based contributors. And as it happens, Абергавенни comes from Abergavenny rather than Y Fenni, showing that some discernment was applied. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
I don't think that's a good idea to try to solve the problems we're faced with by our current OSM data model by setting up a second database. We need an improved data model which fits our needs. Am 29.05.2015 um 11:28 schrieb Jo: We need our own OSMdata instance in between to describe real world objects and concepts. And maybe we could even solve the ridiculous amount of duplication we're experiencing at the moment. True, the editor software will have to be adapted to cope with merges and splits, so the human editor can decide what OSM object(s) belong to what real world object(s). Either that or we should start using relations more intelligently. But they are heavyweight and supposedly they are also complicated. Polyglot 2015-05-29 10:58 GMT+02:00 Martin Koppenhoefer dieterdre...@gmail.com mailto:dieterdre...@gmail.com: 2015-05-28 23:00 GMT+02:00 Andy Mabbett a...@pigsonthewing.org.uk mailto:a...@pigsonthewing.org.uk: On 28 May 2015 at 09:50, Martin Koppenhoefer dieterdre...@gmail.com mailto:dieterdre...@gmail.com wrote: e.g. the en:Spanish Steps / de:Spanische Treppe are called Scalinata di Trinità dei Monti in the local language (it is located at piazza di Spagna, that's where the foreign name comes from, while in Italian it is called after to church it leads to). Naturally, OSM has the original name of this world famous monument, but Wikidata hasn't. It does now. OK, this is one point for you, but it also proves my point: wikidata at the moment is not sufficiently mature (IMHO) to replace name tags in different languages in OSM. Of course you can fix wikidata issues (if you understand how it is done, I haven't had enough time yet to understand how to make edits like this, and the fact that not all tags are shown to me (e.g. only 4 out of all language labels, after I have explicitly clicked on in more languages)) doesn't help. // sidenote: Now I could link the wikidata object of the spanish steps to the OSM object and get an Italian name. But I will not have an Italian wikipedia article about it, because it is covered in the spanish square (piazza di spagna) article in Italian. How would I ideally procede now? a) in wikidata link the article of the piazza di spagna (in italian) to the wikidata object about the spanish steps? a2) like a) but link to an anchor: Piazza_di_Spagna#La_scalinata ? b) in wikipedia split the italian wikipedia article in 2, one for the square and one for the steps? c) in osm add an additional tag like wikipedia:it=Piazza_di_Spagna#La_scalinata to the steps object? d) something different... //sidenote off If we were to massively use wikidata _instead of duplicating some details from there also in our db_ we would have to improve wikidata as well, You'd be welcome to do so. ... and impose our entity structure on them, Really? Good luck with that. what I meant, and what you do confirm below: if for instance there is an object in wikidata which is an administrative entity and a geographic place at the same time, but for OSM we'd need 2 distinct objects, we will have to split the wikidata object. This could be done only if there wasn't resistance from other wikidata users who might want to keep the current unmodified object because it links better to wikipedia articles. We might introduce another object that linked the split objects onto one, which could serve for wikipedia articles, etc. but this is a much more complicated procedure than changing tags in OSM alone. We've always said that we wanted editing to be simple, so that we can maximize the amount of available editors, but with the tight integration of another dynamic dataset (for one of the core competences we are dealing with: toponyms) or it won't work in some cases (and if it doesn't work in some case, it doesn't work at all). That is, of course, nonsense. OK, let's say it is nonesense, because you can accept that a solution works for most of the cases and try work around those that don't work. Currently (all names in OSM) we don't have these problems though. Another issue I see with wikidata is that it contains information and details about spatial objects, but it doesn't contain the geometry it refers to. The geometry is in OSM, is it not? Why would Wikidata want to replicate that? IMHO you have to understand to which geometry you are referring when you make edits, or you might break stuff without noticing it. Wikidata editors would have to look at OSM geometries to ensure that their edit maintains consistency, and OSM users would have to
Re: [OSM-talk] Can wikidata links help fight name inflation?
On Fri May 29 10:42:27 2015 GMT+0100, Nick Whitelegg wrote: Exactly. Us Western Europeans would find Romanised versions of names useful when travelling out of Western Europe (to give a real example: I'm visiting Greece this summer, and while I'm just about at the stage where I think I can decode the Greek alphabet, Romanised versions are definitely helpful), and the converse should also be true. I remember street names in Athens were in both latin and Greek alphabets when I was there, although greek is at least recognisable to most westerners due to its use in maths and science. You don't even need on the ground evidence. You just need someone with knowledge of Cyrillic and Roman alphabets to be able to transliterate Abergavenny into the Cyrillic, presumably. Transliteration is something that can be done at application level, and a traveller would have learned the basics. You still need to be able to check the street names against.what is on the sign. Then why not have a single transliteration? A single cryllic to latin transliteration will serve all languages using the latin alphabet, do we need separate Russian, Ukrainian, Serbo Croat tags when they are identical? Phil (trigpoint) Nick From: Andrew Hain andrewhain...@hotmail.co.uk Sent: 29 May 2015 07:07 To: talk@openstreetmap.org Subject: Re: [OSM-talk] Can wikidata links help fight name inflation? Dave Corley davecorley at gmail.com writes: Lastly, and I think this is important point. To quote the wiki header . the project that creates and distributes free geographic data for the world. Either this is a database of worldwide geodata or its not. There's no half-way in that statement. Either all cultures, languages, countries, people and the variety these elements bring in terms of tagging, is accepted on a universal basis or its not. Thank you Dave. As a British mapper I am ashamed that some people want to make the map of my country less useful, and not only to Russian speakers a long way away. -- Andrew ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk -- Sent from my Jolla ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 29/05/2015, Martin Koppenhoefer dieterdre...@gmail.com wrote: TL;DR; wikidata is a gorgeous project, combining their knowledge with ours is very promising. Still, in my opinion, for the current state of where they are (and where the tools to combine both are), I would _not remove tags_ from OSM just because the same information might be available in wikidata. Agreed fully. It's not either or but and. And the first step is tooling : until I see a map that renders names in my language taken from wikidata (at scale please, the qLabel demo is not good enough) or that points me to the wikipedia article in my language, I won't bother investing time in editing Wikidata for mapping purposes. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
The Ukrainian names were added after the mapper engaged the UK community. Phil (trigpoint) On Fri May 29 11:27:41 2015 GMT+0100, moltonel 3x Combo wrote: On 29/05/2015, SomeoneElse li...@atownsend.org.uk wrote: On 29/05/2015 07:07, Andrew Hain wrote: Thank you Dave. As a British mapper I am ashamed that some people want to make the map of my country less useful, and not only to Russian speakers a long way away. Hang on, where is _anybody_ saying that? The whole point of the thread is about whether it would be possible to use wikidata to make the map _more_ useful to people who don't speak a particular language, not less. The thread opening was indeed about using wikidata translations (which would make the map more usefull for all) and perhaps deprecating name:CC in OSM as a result (which is much more debatable). But then the question came up (By Richard I think ?) of which name:CC should be kept for a particular object, and the large number of name:ru in the UK (reverted by you and others) was given as an example of an unaccepable name:CC. And then some agreed that name:en in non-English-speaking coutries should get the same treatment. This second paragraph can be seen as making the map less useful (I share that POV). It has been argued that the reverts were justified because the names were just transliterations and not real names, but I think that this was overzealous (it wasn't blind tranliteration, there was human skill involved). Even if it was justified, the reversals were either sloppy or biased : see http://www.openstreetmap.org/node/10021976/history which got name:ru Лестер deleted but was allowed to keep name:ukЛестер. And just to be clear: pushing the decision of whether Лестер is the correct russian name of Leicester to wikidata just so that OSM looks cleaner is not a solution. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk -- Sent from my Jolla ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
2015-05-29 11:41 GMT+02:00 Andy Mabbett a...@pigsonthewing.org.uk: On 29 May 2015 at 09:57, Arch Arch 7h3.a...@gmail.com wrote: If the foreign name differs from the transliteration I think no one would oppose to add the name into the OSM database. Then you haven't been paying attention to this thread. The problem is that it's not a transliteration of the characters, that would be easy. It's a transliteration of a pronunciation with characters that happen to be familiar to a person speaking one of many languages which use Latin script. Polyglot ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 29/05/2015, Philip Barnes p...@trigpoint.me.uk wrote: On Fri, 2015-05-29 at 13:14 +0200, moltonel 3x Combo wrote: And as it happens, Абергавенни comes from Abergavenny rather than Y Fenni, showing that some discernment was applied. Not sure I understand that statement, transliterating Y Fenni is equally valid in my view. Precisely, transliterating Y Fenni would have been equaly easy, but the contributor chose the transliteration of Abergavenny instead to use in Russian. That choice was made by a human. That brings Абергавенни closer to being an actual name rather than an automatic transliteration. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 2015-05-29 14:05, SomeoneElse wrote: On 29/05/2015 12:51, Maarten Deen wrote: It depends on what you want. When someone asks me to navigate to Natanzon, Haifa how can I enter it when the only name in the map is נתנזון, חיפה? I don't see why the transliterated name is not important. No-one is saying that the transliterated name is not useful for all sorts of reasons (the thread title can wikidata links help... makes it clear that this is about trying to make it easier to get to these names, not harder). The question here is, in a case when a name:xx isn't widely used in the place and doesn't appear on signs*, how can a user of the data know that they've got there or not? The problem is that there is a third way of writing a place name. Namely the internationalized name of the place in the alphabet that you use. Beijing is Peking in Dutch, but we commonly use Beijing these days. Other countries (Germany, France) are still more prone to using their transliteration. For example: I do not want to see a map that shows name:nl for every place. Some are very obscure and may not be understood, others are more common but are just not used in normal life. I want to see name, but for non-Latin alphabets I probably want name:en. But it should not be a big challenge to identify the character set that's used in a string, so it should also not be very difficult to know when to show name and when to show name:en. On the other hand: for showing a map on my navigation device, I want to see what is on the signposts. So it all depends on the usage. OSM shouldn't be it's own parallel universe - we map what's on the ground. Am I correct in thinking that your idea is not to have things that are not on the ground in OSM? The problem with this is: to have translations by having to look up every name on an external website is not a very workable solution. That means having to load the complete database of that website before you do anything, or accept massive latencies for each lookup. And the relation that is used to lookup the data in this external database may break. It is much more convenient to have name:xx in OSM. Maarten ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 29/05/2015 14:14, moltonel 3x Combo wrote: I'm not a fan of (un)signed=* (the most common in tagginfo) because their meaning is not obvious enough. name:signed=en;cy would be a first and open the multiple-value can of worms. But name:signed=yes/no has 161 uses in taginfo, and name:CC:signed=yes/no seems like an obvious extension. The thread was about trying to avoid name inflation; I was trying to also avoid name:signed inflation :) Cheers, Andy ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 29/05/2015, SomeoneElse li...@atownsend.org.uk wrote: I've used name:signed=no (though this is by no means an accepted tag, and if anyone can come up with a more accepted version that does the same job I'm all ears). Maybe something like name:signed=en;cy might solve the name verifiability problem for Abergavenny? I'll keep those in mind next time I survey such an area. I'm not a fan of (un)signed=* (the most common in tagginfo) because their meaning is not obvious enough. name:signed=en;cy would be a first and open the multiple-value can of worms. But name:signed=yes/no has 161 uses in taginfo, and name:CC:signed=yes/no seems like an obvious extension. What's wrong with name ? What's the UK policy on the content of name for places with Welsh and English names ? If you want to see Welsh names as often as possible but still make the local name more prominent, use local name (welsh name if different) in your map generating script. The problem here is that there are two* equally valid and correct names (in on-the-ground verifiable terms) for Abergavenny. Some communities pick one of the names for the name tag and others put both in. In either case, it should be sufficient for following directions. In this case we have name=Abergavenny, so local mappers favored the english name. While it's not necessary for following directions (use name for that), I agree that it would be nice to know which languages are signposted and verifyable on the ground. IMHO a signpost has never been a requirement for name:CC (or even name actually). Which leads to the name:signed discussion above. Would that cover all the usecases and make people happy ? ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
You don't even need on the ground evidence. You just need someone with knowledge of Cyrillic and Roman alphabets to be able to transliterate Abergavenny into the Cyrillic, presumably. Transliteration is something that can be done at application level, and a traveller would have learned the basics. You still need to be able to check the street names against.what is on the sign. Then why not have a single transliteration? A single cryllic to latin transliteration will serve all languages using the latin alphabet, do we need separate Russian, Ukrainian, Serbo Croat tags when they are identical? Problem is, they are not identical. Russian, Serbian, Ukrainian, Belarusian and other slavic languages have different alphabets, same way as Spanish, German and Norwegian languages all have different letters. Nowa Szkocja and Nova Scotia both sound the same, but are in different languages. Same rules work across all the languages, just - sometimes - they happen to look the same. In turn, looking the same does not guarantee that these names sound the same. What exactly are we trying to save by omitting these locales, what wasn't eaten already by source= on French buildings? :) -- Darafei Komяpa Praliaskouski OSM BY Team - http://openstreetmap.by/ xmpp:m...@komzpa.net mailto:m...@komzpa.net ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
W dniu 29.05.2015 18:43, Andrew Guertin napisał(a): I can imagine a world where no information about businesses is stored in OSM. OSM has a geometry and a wikidata link. Wikidata says what kind of business it is, what its name is, what its contact info it, what its opening hours are, etc. +1 - that would be even better than just a place for categories and other objects relations. Real database instead of flat tags - yummy! -- The train is always on time / The trick is to be ready to put your bags down [A. Cohen] ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
I don't think Wikidata will deem all businesses in the world noteworthy enough for inclusion. In our own instance of OSMdata we can include whatever we like, of course. businesses addresses as first class citizens streets PT stops, routes and we may even be able to figure out a way to include time tables the sky is the limit Jo 2015-05-29 18:43 GMT+02:00 Andrew Guertin andrew.guer...@uvm.edu: On 05/27/2015 05:13 PM, Frederik Ramm wrote: We are a database of geodata [...] Would it not be better to record the wikidata link for London, and then (perhaps in co-operation with people at Wikidata) provide means for people doing map rendering to join OSM data with a separately-loaded translation table from Wikidata? Would this be restricted to just names? I can imagine a world where no information about businesses is stored in OSM. OSM has a geometry and a wikidata link. Wikidata says what kind of business it is, what its name is, what its contact info it, what its opening hours are, etc. That would be a very different world from the one we live in. In lieu of listing them all out, I will just say I can see many benefits to living in that world, and many benefits to what we have now. Is that a goal of this integration? I've been thinking about this for a while, so I have a lot of questions based on the answer... --Andrew ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 05/27/2015 05:13 PM, Frederik Ramm wrote: We are a database of geodata [...] Would it not be better to record the wikidata link for London, and then (perhaps in co-operation with people at Wikidata) provide means for people doing map rendering to join OSM data with a separately-loaded translation table from Wikidata? Would this be restricted to just names? I can imagine a world where no information about businesses is stored in OSM. OSM has a geometry and a wikidata link. Wikidata says what kind of business it is, what its name is, what its contact info it, what its opening hours are, etc. That would be a very different world from the one we live in. In lieu of listing them all out, I will just say I can see many benefits to living in that world, and many benefits to what we have now. Is that a goal of this integration? I've been thinking about this for a while, so I have a lot of questions based on the answer... --Andrew ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 05/29/2015 08:18 AM, Philip Barnes wrote: On Fri, 2015-05-29 at 13:14 +0200, moltonel 3x Combo wrote: And as it happens, Абергавенни comes from Abergavenny rather than Y Fenni, showing that some discernment was applied. Not sure I understand that statement, transliterating Y Fenni is equally valid in my view. Russian speakers have been calling that city Абергавенни for over 150 years. If someone transliterated Y Fenni and said that was the Russian name for the city, they would be wrong. --Andrew ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Andy wrote: For example, when processing OSM data for my own use I'll try and drop unsigned names and refs from roads (there's no point in saying turn left on Foo Street if Foo Street does not appear on the sign). I hope you've not deleted any of the data I added. Just because it doesn't have a sign doesn't mean it doesn't have a name. Approaches I've used to find out its name is to ask a local (usually retired, long standing members of the community are good for this) and researching via council files/public records. Removing names just because they are not on the ground is taking the rule too far and wastes the effort people have put it to the map. It is tagging for the routing engine - we should adopt another method to stop routing engines announcing turn left to xxx where xxx is unsigned. Example a signed:name=no tag. Rob ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
pet, 29. svi 2015. 18:45 Andrew Guertin andrew.guer...@uvm.edu je napisao: On 05/27/2015 05:13 PM, Frederik Ramm wrote: We are a database of geodata [...] Would it not be better to record the wikidata link for London, and then (perhaps in co-operation with people at Wikidata) provide means for people doing map rendering to join OSM data with a separately-loaded translation table from Wikidata? Would this be restricted to just names? I can imagine a world where no information about businesses is stored in OSM. OSM has a geometry and a wikidata link. Wikidata says what kind of business it is, what its name is, what its contact info it, what its opening hours are, etc. That would be a very different world from the one we live in. In lieu of listing them all out, I will just say I can see many benefits to living in that world, and many benefits to what we have now. Is that a goal of this integration? I've been thinking about this for a while, so I have a lot of questions based on the answer... --Andrew ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk What I'm sure about is that Wikidata isn't the place for opening hours of shops and businesses. The question is, should we build our own OSMData or OpenPOIData that has that information, or is OSM good enough. Janko ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 29/05/15 18:18, Janko Mihelić wrote: What I'm sure about is that Wikidata isn't the place for opening hours of shops and businesses. The question is, should we build our own OSMData or OpenPOIData that has that information, or is OSM good enough. Or simply provide a well documented mechanism by which third parties can provided secondary layers of data! But there is a difference between the accessing of names such as provided by geonames, and allowing that lookup to happen in any language and adding what really should be 'cloud sourced' data such as the opening time, menu, drinks list etc of a business at x,y. Further providing a google style translation of that again is something that one would not expect in OSM? Although a tool that allows a particular language version of pop-up data attached to OSM may be appropriate going forwards. If we have a correct website link for the business, there is no need to store opening times IN OSM ... but an historic view may well require archiving older business material ... which is where OHM starts plugging the gaps. Again ... do we need ever translation of every object in OSM to be served up when all we actually need is our own language, and perhaps the native languages of the target object. OSMData which provides a filtered view of the available data is what we are talking about, and while a full XML planet dump with every possible translation is who the system currently works, relational data bases are the logical way to provided filtered live data? Should that layer be provided by a third party? I think the answer there is probably no, but providing an interface where third party versions of the data can be used makes sense. -- Lester Caine - G8HFL - Contact - http://lsces.co.uk/wiki/?page=contact L.S.Caine Electronic Services - http://lsces.co.uk EnquirySolve - http://enquirysolve.com/ Model Engineers Digital Workshop - http://medw.co.uk Rainbow Digital Media - http://rainbowdigitalmedia.co.uk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
2015-05-29 0:50 GMT+03:00 SomeoneElse li...@atownsend.org.uk: On 28/05/2015 22:27, moltonel 3x Combo wrote: You can argue against machine-made non-reviewed translitterations, because they don't add anything that a data consumer couldn't and because they likely contain mistakes. But that's apparenlty not the case of the name:ru changeset that got reverted. The source of the names in the reverted changeset is described in the changeset discussion http://www.openstreetmap.org/changeset/30451655 : to find russian names for places, I googled for English place names plus достопримечательности, погода or some other russian word. Looks like you completely miss the point I'm native Russian speaker who also speaks English a little bit. I know how a russian would call that place the minute I see it. Googling is a way to cross-check if my guess is valid - the corner case about Slough was given a couple of letters above. Some places with unusual pronounciation got their name:ru 5 years ago - why weren't these reverted since then? http://www.openstreetmap.org/node/21280555/history Some live happily with name:he, name:lt and even name:ja, but name:ru was removed from them repeatedly: http://www.openstreetmap.org/node/10021976/history I'm sorry to ask - isn't japanese tag there a transcription to hieroglyphs? When I ask google translate to pronounce it, I hear what I would write in cyrillic as Лестаар. So, why was it kept? I also see links to Name page referring IRC discussions. http://wiki.openstreetmap.org/w/index.php?title=Namesdiff=nextoldid=976484 While IRC discussions may be good for intra-cultural things, it's better to discuss such large topics on mailing lists, and at least not revert such data silently. -- Darafei Komяpa Praliaskouski OSM BY Team - http://openstreetmap.by/ xmpp:m...@komzpa.net mailto:m...@komzpa.net ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On Thu, May 28, 2015 at 1:27 PM, Dave Corley davecor...@gmail.com wrote: On the first point, why not? There are maps of the world in English, French, German etc etc. I see no logical reason to object to certain languages being used in the name tag. That is the whole point of the flexibility of that tag. The scale of usage is null and void. Either its acceptable everywhere or its acceptable nowhere. Anything in between is entirely subjective and completely unfair. On the second point, since when do we care about the motivation about why certain data gets added. If someone wants to add 3d tagging to utilise in a 3d map they are creating, do we nitpick about that? If Scout adds speed limits to make their app more effective, do we complain? Lastly, and I think this is important point. To quote the wiki header . the project that creates and distributes free geographic data for the world. Either this is a database of worldwide geodata or its not. There's no half-way in that statement. Either all cultures, languages, countries, people and the variety these elements bring in terms of tagging, is accepted on a universal basis or its not. If its not, then that's an entirely different conversation. +1 -- @osm_seattle osm_seattle.snowandsnow.us OpenStreetMap: Maps with a human touch ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Hi, On 05/28/2015 09:56 PM, Jóhannes Birgir Jensson wrote: What I would not support in OSM, and like to outsource to Wikidata or other, is if speakers of these 20 languages were to start assigning name tags in their language to thousands of places in, say, the UK. Where do you draw the limit? Does that mean you agree there should be a limit, or would you allow each and every of the several thousand languages on the planet to add their name tag to the London node? How would disputes be handled? name:en is on thousands of nodes but in many of these places it is not an official or even a minority language but an extra language. Yes, I've thought about that; name:en is very useful for me but ultimately, if the locals don't use it, then it isn't on the ground, and then it shouldn't be in OSM really. Should languages with over 1 billion (native and non-native) speakers be allowed to put name:xx all over the world. 500m? 10m? 10 thousand? I don't think any size gives one the right to cultural imperialism; let the locals decide what names *they* think they have and that's it. Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
+1 to Dave. OSM is universal, full stop. On Thu, May 28, 2015 at 4:27 PM, Dave Corley davecor...@gmail.com wrote: I am honestly stunned this thread has gone on for as long as it has. In regards to On 05/28/2015 09:19 PM, Frederik Ramm wrote: What I would not support in OSM, and like to outsource to Wikidata or other, is if speakers of these 20 languages were to start assigning name tags in their language to thousands of places in, say, the UK. *Especially* if their reasoning was that this makes it nicer for them to run a tank through these places in their own-language war simulation with their buddies. On the first point, why not? There are maps of the world in English, French, German etc etc. I see no logical reason to object to certain languages being used in the name tag. That is the whole point of the flexibility of that tag. The scale of usage is null and void. Either its acceptable everywhere or its acceptable nowhere. Anything in between is entirely subjective and completely unfair. On the second point, since when do we care about the motivation about why certain data gets added. If someone wants to add 3d tagging to utilise in a 3d map they are creating, do we nitpick about that? If Scout adds speed limits to make their app more effective, do we complain? Lastly, and I think this is important point. To quote the wiki header . the project that creates and distributes free geographic data for the world. Either this is a database of worldwide geodata or its not. There's no half-way in that statement. Either all cultures, languages, countries, people and the variety these elements bring in terms of tagging, is accepted on a universal basis or its not. If its not, then that's an entirely different conversation. Dave ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 28 May 2015 at 11:09, Martin Koppenhoefer dieterdre...@gmail.com wrote: 2015-05-28 11:41 GMT+02:00 Frederik Ramm frede...@remote.org: But let's not get sidetracked, that's a different discussion from the Wikidata question. I just hope that Wikidata doesn't list New Brige as the English name of Pont Neuf or else they have a problem ;) actually they do for Spanish and Catalan: www.wikidata.org/wiki/Q335277 I guess they do have a problem now? Or perhaps that's what people speaking those languages actually call the bridge? Another issue that can be seen here: nobody will want this Puente Nuevo (París) as a label for a bridge on a map (París) Seen where? Every single entity in wikidata I have looked at had some issues in one or the other way, Either your sampling is flawed or you're exaggerating. if you open up this can of worms for names, there will soon be the desire to extend the concept This is not a can of worms; but Wikidata does offer far more than names. -- Andy Mabbett @pigsonthewing http://pigsonthewing.org.uk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 28/05/2015, Andy Mabbett a...@pigsonthewing.org.uk wrote: https://www.wikidata.org/wiki/Q84 You can see its names in various languages by clicking the Labels list tab (then note slidebar) Call me stupid, but I don't see any clickable label list tab ? Sorry; you won't see it if you're not logged in. Ah, I thought for a moment that I'd need to create an account and log in, but then I saw that anonymous edits were accepted, so I assumed that I could see everything that was to be seen. Then I saw the translations in the json output that somebody linked, and wondered again if I was blind when reading the renderd html. You'd need to log in and enable this gadget: https://www.wikidata.org/wiki/Wikidata:Tools/Gadgets#LabelLister I'll ask if there's a method for non-logged-in users to see them. Please do. I'm really surprised that translations aren't visible like every other properties. Is there a technical or editorial reason ? ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Am 28.05.2015 um 22:20 schrieb Frederik Ramm frede...@remote.org: Yes, I've thought about that; name:en is very useful for me but ultimately, if the locals don't use it, then it isn't on the ground, and then it shouldn't be in OSM really. are we really insisting in locals, or can a traveller be on the ground as well? This traveller could communicate with other travelers... cheers Martin ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 28/05/2015, Andrew Guertin andrew.guer...@uvm.edu wrote: Considering the existence of the former Soviet Union, and especially that there are areas of Ukraine where both Russian and Ukrainian are spoken and most roads, places, etc have names (and thus tags) in both languages, this number of 582,653 name:ru tags is hard to interpret. My skill with overpass-turbo isn't the best, but I was able to relatively easily limit a search to a bounding box around North and South America. Within that box, a search returned 2648 nodes and 909 ways with name:ru (relations timed out). Taginfo offers a quick way to view this: http://taginfo.osm.org/search?q=name http://taginfo.osm.org/keys/name%3Aen#map http://taginfo.osm.org/keys/name%3Aru#map http://taginfo.osm.org/keys/int_name#map http://taginfo.osm.org/keys/name%3Aja#map http://taginfo.osm.org/keys/name%3Ade#map http://taginfo.osm.org/keys/name%3Afr#map http://taginfo.osm.org/keys/name%3Aar#map ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Þann 28.5.2015 19:43, skrifaði Frederik Ramm: Hi, On 05/28/2015 09:19 PM, Frederik Ramm wrote: What I would not support in OSM, and like to outsource to Wikidata or other, is if speakers of these 20 languages were to start assigning name tags in their language to thousands of places in, say, the UK. *Especially* if their reasoning was that this makes it nicer for them to run a tank through these places in their own-language war simulation with their buddies. Disclaimer, I own some of the tanks that might run through some of the places. As any consumer I've been critical of their work and loved and hated it with equal passion, depending on what they did in the last patch release. Having said that the one nice thing about open projects is the added value people put into them and get out of them for a myriad of reasons. A gaming company adding information (yes it is information) to the map can be beneficial to another consumer, for example a Russian style TripAdvisor or something completely different. Is a gaming company not worthy of adding information? Again a matter of where do you draw the line. They are not inventing new towns or places, they are not creating battlefields. Do we now require each contributor to disclose the reason for why they are adding information to the map? This revert and discussions about it should not become too specific to the contributor whose work was reverted. These edits and reverts point at a vital issue as we try to grow the map in areas where hundreds of languages suddenly eye the ability to get maps in their own languages using the OSM data and tools that exist for it. I propose the OSMF tackle these two questions, who should be able to add information to the map and what do we do about languages and where are the limits. --Jói / Stalfur ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On Thu, 2015-05-28 at 22:20 +0200, Frederik Ramm wrote: Yes, I've thought about that; name:en is very useful for me but ultimately, if the locals don't use it, then it isn't on the ground, and then it shouldn't be in OSM really. Absolutely agree, there is a tendency to have name for places that really don't need it. As a traveller I would far rather maps show what I am going to see on the ground, or a roadsign than some made up name. A German city, I think it was Würzburg, was in the news last year. It is a name I am familiar with, however the BBC news report had dug up an English spelling I was totally unaware existed. Many are just plain silly, why does English put an 's' on the end of Lyon and Marseille. Berlin does not need a name;en tag. Phil (trigpoint) ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 28 May 2015 at 10:59, Steve Doerr doerr.step...@gmail.com wrote: There might be a case for adding pronunciations (of 'difficult' names at least) to the OSM database. Someone must have proposed a tagging scheme for this, surely? Yes; wikidata= Wikidata will have both IPA [1] phonetic guides to pronunciation, and in some cases links to audio files. [1] https://en.wikipedia.org/wiki/Help:IPA_for_English - as used in Wikipedias. -- Andy Mabbett @pigsonthewing http://pigsonthewing.org.uk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 28/05/2015 22:27, moltonel 3x Combo wrote: You can argue against machine-made non-reviewed translitterations, because they don't add anything that a data consumer couldn't and because they likely contain mistakes. But that's apparenlty not the case of the name:ru changeset that got reverted. The source of the names in the reverted changeset is described in the changeset discussion http://www.openstreetmap.org/changeset/30451655 : to find russian names for places, I googled for English place names plus достопримечательности, погода or some other russian word. Cheers, Andy ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 05/28/2015 03:43 PM, Frederik Ramm wrote: *Especially* if their reasoning was that this makes it nicer for them to run a tank through these places in their own-language war simulation with their buddies. If the data is valid, it doesn't matter what the use case is. I'm HAPPY knowing that people are using the data I contribute in all sorts of ways and contributing back so I can use theirs. --Andrew ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Þann 28.5.2015 19:19, skrifaði Frederik Ramm: What I would not support in OSM, and like to outsource to Wikidata or other, is if speakers of these 20 languages were to start assigning name tags in their language to thousands of places in, say, the UK. Bye Frederik Where do you draw the limit? name:en is on thousands of nodes but in many of these places it is not an official or even a minority language but an extra language. Should languages with over 1 billion (native and non-native) speakers be allowed to put name:xx all over the world. 500m? 10m? 10 thousand? --Jói / Stalfur ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
I am honestly stunned this thread has gone on for as long as it has. In regards to On 05/28/2015 09:19 PM, Frederik Ramm wrote: What I would not support in OSM, and like to outsource to Wikidata or other, is if speakers of these 20 languages were to start assigning name tags in their language to thousands of places in, say, the UK. *Especially* if their reasoning was that this makes it nicer for them to run a tank through these places in their own-language war simulation with their buddies. On the first point, why not? There are maps of the world in English, French, German etc etc. I see no logical reason to object to certain languages being used in the name tag. That is the whole point of the flexibility of that tag. The scale of usage is null and void. Either its acceptable everywhere or its acceptable nowhere. Anything in between is entirely subjective and completely unfair. On the second point, since when do we care about the motivation about why certain data gets added. If someone wants to add 3d tagging to utilise in a 3d map they are creating, do we nitpick about that? If Scout adds speed limits to make their app more effective, do we complain? Lastly, and I think this is important point. To quote the wiki header . the project that creates and distributes free geographic data for the world. Either this is a database of worldwide geodata or its not. There's no half-way in that statement. Either all cultures, languages, countries, people and the variety these elements bring in terms of tagging, is accepted on a universal basis or its not. If its not, then that's an entirely different conversation. Dave ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
I agree with Dave here, but add some general remarks : Please handle the questions of should FOO-language name of an object be allowed in the database ? and should that databse be OSM or Wikidata ? separately. The decision of whether Абергавенни should be recorded as the Russian name for Abergavenny should be the same for Wikidata and OSM. I disagree with the idea that only local languages are acceptable, firstly because you don't know that there isn't a local speaking that language natively, and secondly because people have given names to foreign places before even the first maps were drawn. A name doesn't have to go through a vetting process to become a name. It becomes one as soon as it is used and recognised by more than one person. There's no point in complaining about name:ru (580k) when name:en is at 1320k, name:ja 320k, name:de 280k, name:fr 270k, and the taginfo map for all of those shows a worldwide distribution roughly matching the worldwide OSM data density. You can argue against machine-made non-reviewed translitterations, because they don't add anything that a data consumer couldn't and because they likely contain mistakes. But that's apparenlty not the case of the name:ru changeset that got reverted. I'm not worried about all named osm objects someday getting thousands of name:CC tags because realistically that's not going to happen. London still has a measly 154. Europe 46. Worldwide interest will decide how far a given name goes, but I'd be surprised (and pleased) if one place ever goes above 500. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On Thursday 28 May 2015, Andrew Guertin wrote: There is a fundamental difference between an actual name for a place and a translation of one of those names I DO agree with this statement[1]. However, I think that the point at which a word stops being a transliteration and starts being a native word is much sooner than you seem to. I'm not a linguist, but if I had to pin down when I think a word becomes part of a language, I'd say whenever the person using it doesn't think they're code-switching.[2] It's clear that in many cases, the people writing Абергавенни don't consider themselves to ever be switching out of Russian. To me, that makes Абергавенни an actual name for the place. That might be the case but for OSM what matters should be if the use of this reference is sufficiently widespread, specific and prominent to be verifiable for mappers. If i use a certain 'name' to refer to a certain geographic feature this is not yet a name of that feature in any wider sense no matter how i generated this name. When i start communicating that name by refering to this feature with that name to others and others pick up the name and use it themselves this changes. When this use of the name i chose gets widespread - both in absolute numbers, i.e. people actually using it frequently and in relative numbers, i.e. this name becomes the dominating name to refer to that feature within a certain context, like in a certain language, it will be sufficiently established to be tagged in OSM (in my opinion - others might see this differently). Note the key part here is the conscious and specific use of the name for the feature in question. Automatically generated lists and databases do not count. These might influence peoples' use of names but do not establish a name on their own. -- Christoph Hormann http://www.imagico.de/ ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 28 May 2015 at 09:50, Martin Koppenhoefer dieterdre...@gmail.com wrote: e.g. the en:Spanish Steps / de:Spanische Treppe are called Scalinata di Trinità dei Monti in the local language (it is located at piazza di Spagna, that's where the foreign name comes from, while in Italian it is called after to church it leads to). Naturally, OSM has the original name of this world famous monument, but Wikidata hasn't. It does now. Wait, it hasn't the original name of this three-star-tourist-attraction, how's that? Have a look here: http://www.wikidata.org/wiki/Q848072 the reason is that the Italian wikipedia hasn't got an article about the steps No, that so not the reason, The reason was simply that no one had got around to adding it. Rather like OSM not being complete, yet. If we were to massively use wikidata _instead of duplicating some details from there also in our db_ we would have to improve wikidata as well, You'd be welcome to do so. Just as Wikipedia editors who want better maps for Wikipedia will improve OSM. and impose our entity structure on them, Really? Good luck with that. But why do you think you would need to? or it won't work in some cases (and if it doesn't work in some case, it doesn't work at all). That is, of course, nonsense. Another issue I see with wikidata is that it contains information and details about spatial objects, but it doesn't contain the geometry it refers to. The geometry is in OSM, is it not? Why would Wikidata want to replicate that? Have a look at the Berlin object: http://www.wikidata.org/wiki/Q64 This covers both, an administrative entity and a geographic place in one object (no problem here, but can be a problem elsewhere). Eventually, these will be split. This object has a property instance of metropolis http://www.wikidata.org/wiki/Q200250 I don't want to discuss whether Berlin is a metropolis or not, what I want to point out is that there seem to be different criteria defined for different languages: These descriptions aid users; they are not proscriptive. There are also local and cultural variations. Just like city in OSM. -- Andy Mabbett @pigsonthewing http://pigsonthewing.org.uk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 28 May 2015 at 10:41, Frederik Ramm frede...@remote.org wrote: I just hope that Wikidata doesn't list New Brige as the English name of Pont Neuf or else they have a problem ;) It alls it Pont neuf: https://www.wikidata.org/wiki/Q335277 and has a property that tells you that that is its official name, which is in the French language. At some point, it will have a property that tells you that the name /means/ New Bridge. It is my impression that a large proportion of name:xx tags in OSM are added by naming specialists who do little else than large scale name additions; it would probably not be too much to ask for them to indulge Wikidata instead of OSM. Please do! if a mapper is of the opinion that no matching Wikidata object exists for an OSM feature, then they shouldn't use a wikidata tag, that much is clear! Or they could, if suitable, crate a WIkidata item... -- Andy Mabbett @pigsonthewing http://pigsonthewing.org.uk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 27/05/15 22:13, Frederik Ramm wrote: We could then limit ourselves to using a name tag for the locally used name, or continue to allow a name:xx but only if these languages were actually used by the local population; throw in an int_name if you want (but some may say that's already an unfair privilege for users of English and the Latin alphabet). Anything else - i.e. names used for a place in other languages than the local ones - would be off-topic for OSM and should be recorded in Wikidata. Do you think Wikidata could play that role, and take the burden off of us? Or is Wikidata not mature enough for that yet, or even unsuitable? I've followed the to and fro on this and most things make sense. I would not be happy with a third party providing a 'service' that is currently handled via Nominatim and Geonames but this comes back to a proposal I put forward some ten years ago. At the end of the day, a 'translation' of everything contained in database is needed once one switches to a different native language. English has been adopted as the base language for the tagging, and currently it is expected that each tool creates it's own translations of that for 'local' use, when a central dictionary makes a LOT more sense. Adding 8 million locations each with it's own set of 'translations' dovetails in with that, and tools can then simply return a single language version of the data rather than having to manage hundreds of different versions of each name. If a translation exists in the dictionary, that is used, otherwise a fall back to a secondary language can be used eventually falling back to an English identifier. For other users of the material, they could substitute their own dictionary in place of the 'official' one, so we can have Klingon or Jedi for those using the maps for gaming purposes. But I don't think we can rely on a third party like Wikidata to provide this dictionary ... they may however parallel it's operation? -- Lester Caine - G8HFL - Contact - http://lsces.co.uk/wiki/?page=contact L.S.Caine Electronic Services - http://lsces.co.uk EnquirySolve - http://enquirysolve.com/ Model Engineers Digital Workshop - http://medw.co.uk Rainbow Digital Media - http://rainbowdigitalmedia.co.uk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 28 May 2015 at 13:33, moltonel 3x Combo molto...@gmail.com wrote: On 28/05/2015, Martin Koppenhoefer dieterdre...@gmail.com wrote: Another issue that can be seen here: nobody will want this Puente Nuevo (París) as a label for a bridge on a map (París) Funny ah? Every single entity in wikidata I have looked at had some issues in one or the other way, I believe we would get more problems than we would solve. The issue here is that these strings are the name of the wikipedia article in various language, which is *not* the same as the name of the location in various languages. Wikipedia likes to add some disambiguation text, which we do not want in OSM. Wikidata labels do not include that disambiguation (some exceptions may be found, but they are user errors,and are being weeded out). https://www.wikidata.org/wiki/Q84 You can see its names in various languages by clicking the Labels list tab (then note slidebar) Call me stupid, but I don't see any clickable label list tab ? Sorry; you won't see it if you're not logged in. You'd need to log in and enable this gadget: https://www.wikidata.org/wiki/Wikidata:Tools/Gadgets#LabelLister I'll ask if there's a method for non-logged-in users to see them. The closest thing I see is the list of wikipedia articles, which has the problem mentioned above. Those are not the labels -- Andy Mabbett @pigsonthewing http://pigsonthewing.org.uk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
[Resending, to this list] On 27 May 2015 at 23:03, Alex Barth a...@mapbox.com wrote: OpenStreetMap is the spatial representation of the world - wouldn't it make sense then to also store the translations for locations in OpenStreetMap? No. Wikidata exists because the Wikipedia community realised that it is folly to store and maintain duplicate copies of the same data in more than one location (in this case, Wikipedias in different languages). Not only is that wasteful - especially in terms of volunteer effort - but it also results in de-synchronisation. Instead, Wikidata was created to hold data which would then be used by multiple Wikipedias (and sister Wikimedia projects), and made available to third parties, of which OSM is one. We simply don't have the spare volunteer capacity to needlessly dupliate all that effort; even if it were the sensible thing to do, which it is not. -- Andy Mabbett @pigsonthewing http://pigsonthewing.org.uk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Hi, On 05/28/2015 10:27 PM, Dave Corley wrote: Either this is a database of worldwide geodata or its not. There's no half-way in that statement. Either all cultures, languages, countries, people and the variety these elements bring in terms of tagging, is accepted on a universal basis or its not. It is not. Your either-or is far too simplistic even for today's OpenStreetMap, and sadly doesn't help this discussion at all. There's quite a couple of cultures/world views that we are already rejecting today, for example nationalist world views that would apply certain name tags to objects clearly outside their jurisdiction and against the expressed wish of the local population. Decisions have to be made, especially were one person's freedom starts to impact another person's freedom; it never was free-for-all. Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 28 May 2015 at 09:50, SomeoneElse li...@atownsend.org.uk wrote: On 27/05/2015 22:56, Andy Mabbett wrote: A demonstrator, using Wikidata labels, is: http://googleknowledge.github.io/qlabel/demo/map/ (choose select language). Coders might enjoy viewing the source code. That's interesting, but seems just to do multiple http transactions to get the names it needs (something that's not really scalable). As I said, it's a *demonstrator*. It's not meant to scale. What I'd typically want to do with wikidata would be something like: 1) define a series of properties that I'm interested in. Unfortunately I don't see this in any sort of sensible format - I just see a bunch of web pages like http://www.wikidata.org/wiki/Wikidata:List_of_properties which isn't really helpful. I'm not sure what you're looking for. I suspect that these are problems that someone, somewhere has already solved Indeed. -- Andy Mabbett @pigsonthewing http://pigsonthewing.org.uk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 05/27/2015 05:13 PM, Frederik Ramm wrote: Not only well-known tourist magnets carry foreign names; some dedicated language mappers have gone over and beyond the call of duty and added, for example, name:ru tags even to small villages: http://taginfo.openstreetmap.org/keys/name%3Aru#map (This is a matter currently under investigation by Data Working Group and it is relatively certain that not all 582,653 name:ru tags will remain.) Considering the existence of the former Soviet Union, and especially that there are areas of Ukraine where both Russian and Ukrainian are spoken and most roads, places, etc have names (and thus tags) in both languages, this number of 582,653 name:ru tags is hard to interpret. My skill with overpass-turbo isn't the best, but I was able to relatively easily limit a search to a bounding box around North and South America. Within that box, a search returned 2648 nodes and 909 ways with name:ru (relations timed out). Considering in 2007 Russian was the primary language spoken in the homes of over 850,000 individuals living in the United States[1], 3500 features with Russian names across all of North and South America seems very low, and there's lots of opportunity for more data to be added. --Andrew [1] https://en.wikipedia.org/wiki/Russian_language#Geographic_distribution ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Hi, On 05/28/2015 08:53 PM, Jóhannes Birgir Jensson wrote: As an example Botswana has English as an official language and Setswana as the overwhelming majority language. However it also has 20 smaller languages, some of them shared with neighboring nations and some very small local ones. Those languages already fight for their existence as the English names are given to their villages on official maps, with maybe the local in a footnote. Using OSM we can present them with maps of their own homes in their own languages. That's something I would support. What I would not support in OSM, and like to outsource to Wikidata or other, is if speakers of these 20 languages were to start assigning name tags in their language to thousands of places in, say, the UK. Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
I have one general question. Some OSM relations contain a link to a Wikidata item. For those relations, would it be possible to automatically import the names in several languages and do so in a way that changes in OSM also change Wikidata? Thanks, Micru On Thu, May 28, 2015 at 6:37 PM, SomeoneElse li...@atownsend.org.uk wrote: On 28/05/2015 16:38, Andrew Guertin wrote: A quick internet search shows plenty of results for Абергавенни, including Wikipedia, hotel booking sites, and Harry Potter websites, and by looking at Google's book results, you can see that it's been in use since at least the 1800s. And with just a few minutes' look, I found someone from the next city over using the name[1]. I understand this was just an example, but it seems to show the opposite of what you wanted. The town with the English name Abergavenny also has a Russian name Абергавенни, which is in use by locals, and has been established for hundreds of years. No, it does not. Abergavenny / Y Fenni has actual names that people from there use to describe the place (and appears on signs) in two languages; Абергавенни is merely a translation of one of them. It's not verifiable on the ground. There is a fundamental difference between an actual name for a place and a translation of one of those names - it's that distinction that we would lose by populating name:ru, name:xx or whatever alongside name:cy and name:en. The russian-language link talking about Abergavenny Food Festival does indeed use the word Абергавенни- and that's a translation of Abergavenny in that message (they even put Abergavenny in brackets afterwards to make it clear that that's what it is - it's clearly not guaranteed to be understood on its own). If Абергавенни is added as name:ru for Abergavenny, how would we tell the real names (the ones that people have historically used locally to refer to the place) from the tranlations? Cheers, Andy ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk -- Etiamsi omnes, ego non ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Þann 28.5.2015 15:38, skrifaði Andrew Guertin: While your exact words here aren't wrong, I think you're severely underestimating what objects have names in what languages. Russia and the UK are major world powers that have had a lot of interaction as both allies and enemies, economically, militarily, and culturally, and there are tens to hundreds of thousands of people who were born in Russia living in the UK[2]. It would be pretty absurd to for place names NOT to exist, and as shown above the evidence shows that they do exist and are in use. For that reason I think the revert was wrong, and the edit should be allowed to be re-performed. I agree that the revert was maybe somewhat overzealous. While we do not have a viable and published method and procedures in place on how to handle multi-lingual names in a different manner, e.g. with an outside source, then reverting any name translations is hard to argue for. The multi-lingual aspect of OSM is one of its absolutely strongest advantages. Being able to create maps of regions in the languages of the local population is quite simply fantastical and maybe under-appreciated by those who speak dominant languages, worldwide or regional. As an example Botswana has English as an official language and Setswana as the overwhelming majority language. However it also has 20 smaller languages, some of them shared with neighboring nations and some very small local ones. Those languages already fight for their existence as the English names are given to their villages on official maps, with maybe the local in a footnote. Using OSM we can present them with maps of their own homes in their own languages. So the question is also, do we outsource this strength of OSM? Wikidata has notability problems and so I'm wary of it, no matter the guidelines there is no shortage of people who deem their judgement in what human knowledge is worth retaining to be superior to everything else. Is name inflation bad? Doubtful in my mind, even the opposite. --Jói / Stalfur ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 05/28/2015 12:37 PM, SomeoneElse wrote: There is a fundamental difference between an actual name for a place and a translation of one of those names I DO agree with this statement[1]. However, I think that the point at which a word stops being a transliteration and starts being a native word is much sooner than you seem to. I'm not a linguist, but if I had to pin down when I think a word becomes part of a language, I'd say whenever the person using it doesn't think they're code-switching.[2] It's clear that in many cases, the people writing Абергавенни don't consider themselves to ever be switching out of Russian. To me, that makes Абергавенни an actual name for the place. The town with the English name Abergavenny also has a Russian name Абергавенни, which is in use by locals, and has been established for hundreds of years. No, it does not. Abergavenny / Y Fenni has actual names that people from there use to describe the place (and appears on signs) in two languages; Абергавенни is merely a translation of one of them. It's not verifiable on the ground. The question of actual names versus transliterations is addressed above, but with respect to on the ground, I assert that if you asked a local who spoke Russian the question What is the Russian-language name for this town?, they would reply Абергавенни. Assuming that assertion is correct, is on the ground satisfied? Do you think the assertion is incorrect? --Andrew [1] I think the word translation is wrong here. Translation takes something in one language and expresses the same meaning in another language. Transliteration takes something in one language and expresses the same sounds in another language. [2] Obviously there's a separate barrier for something to be generally accepted rather than just one person's made up word. That barrier is pretty low in the case of a place name, where most people would make up the same new word anyway. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Hi, On 05/28/2015 09:19 PM, Frederik Ramm wrote: What I would not support in OSM, and like to outsource to Wikidata or other, is if speakers of these 20 languages were to start assigning name tags in their language to thousands of places in, say, the UK. *Especially* if their reasoning was that this makes it nicer for them to run a tank through these places in their own-language war simulation with their buddies. Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 2015-05-28 22:20, Frederik Ramm wrote: Hi, On 05/28/2015 09:56 PM, Jóhannes Birgir Jensson wrote: What I would not support in OSM, and like to outsource to Wikidata or other, is if speakers of these 20 languages were to start assigning name tags in their language to thousands of places in, say, the UK. Where do you draw the limit? Does that mean you agree there should be a limit, or would you allow each and every of the several thousand languages on the planet to add their name tag to the London node? How would disputes be handled? name:en is on thousands of nodes but in many of these places it is not an official or even a minority language but an extra language. Yes, I've thought about that; name:en is very useful for me but ultimately, if the locals don't use it, then it isn't on the ground, and then it shouldn't be in OSM really. So, the map should only be useful to people living in the vicinity of certain places? Like my previously issue, countries which use a different alphabet then yours. Place names in China are ususally only in Chinese, so I should not be able to read them on the map? I say again: the fact that I can not read names in China, Russia, Laos, Israel, most arabic countries is one of my biggest gripes with the map. Why should I not be able to use the map in thos countries and why should people from those countries who probably may not all have a good understanding of the Latin alphabet not be able to use the map where the alphabet is Latin? For that you need transliterated names which probably are not used by locals and are usually not on local street signs so not verifiable. Maarten ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 27/05/2015 22:56, Andy Mabbett wrote: A demonstrator, using Wikidata labels, is: http://googleknowledge.github.io/qlabel/demo/map/ (choose select language). Coders might enjoy viewing the source code. That's interesting, but seems just to do multiple http transactions to get the names it needs (something that's not really scalable). What I'd typically want to do with wikidata would be something like: 1) define a series of properties that I'm interested in. 2) extract that information from wikidata (either a structured download or from some sort of dump) in one go, not as a series of http transactions. 3) load that into local database tables where it can be easily accessed. (1) might be something like villages in Derbyshire or mills in the Derwent valley or something broader (suppose I wanted to include who owns what building in a database containing OSM data). Unfortunately I don't see this in any sort of sensible format - I just see a bunch of web pages like http://www.wikidata.org/wiki/Wikidata:List_of_properties which isn't really helpful. 2) must be a problem that people have solved already As ever, stackoverflow has some of the answers but some of the questions such as http://stackoverflow.com/questions/28391434/how-to-parse-bigdata-json-file-wikidata-in-c-efficiently suggest to me I really wouldn't start from there if I were you (though http://stackoverflow.com/questions/29886388/get-all-wikidata-items-that-are-an-instance-of-a-given-item is closer). I suspect that these are problems that someone, somewhere has already solved, but I'm not seeing obvious answers that aren't a bit of a cludgy hack, or requires something from Google that's going away in 32 days, or whatever. Cheers, Andy ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
But what exactly is the problem that you're trying to solve with this idea? Database size? There are much bigger contributing factors to database size than this, like rampant data redundancy everywhere, botched mechanical edits etc. Complexity of the UI of editors? I'm sure they can manage to optimize it. Paweł On Wed, May 27, 2015, at 23:13, Frederik Ramm wrote: Hi, we're seeing more and more name:xx tags on OSM objects. Not only are speakers of widely used languages adding their language tags all over the world; but rising interest in OSM also brings us to the attention of language lovers and speakers of minority languages. The less established a language is, the more committed its few proponents to have their language respected and recorded. The place node for London has 154 name tags as we speak, but there are several thousand languages in use on the planet, so there's still room for enhancement. Not only well-known tourist magnets carry foreign names; some dedicated language mappers have gone over and beyond the call of duty and added, for example, name:ru tags even to small villages: http://taginfo.openstreetmap.org/keys/name%3Aru#map (This is a matter currently under investigation by Data Working Group and it is relatively certain that not all 582,653 name:ru tags will remain.) It is difficult to judge when such foreign names have a right to be there, and when they're just inventions or name translations or transliterations. I guess we'll have to make rules on that somehow, but at the same time I dread doing it, and I wonder: If a place has a wikidata tag, could/should we then simply defer to Wikidata for names in other languages? We are a database of geodata and not one of international cultural heritage; even if London has a name in over 2000 languages, is OSM really the place to record these 2000 names? Would it not be better to record the wikidata link for London, and then (perhaps in co-operation with people at Wikidata) provide means for people doing map rendering to join OSM data with a separately-loaded translation table from Wikidata? We could then limit ourselves to using a name tag for the locally used name, or continue to allow a name:xx but only if these languages were actually used by the local population; throw in an int_name if you want (but some may say that's already an unfair privilege for users of English and the Latin alphabet). Anything else - i.e. names used for a place in other languages than the local ones - would be off-topic for OSM and should be recorded in Wikidata. Do you think Wikidata could play that role, and take the burden off of us? Or is Wikidata not mature enough for that yet, or even unsuitable? Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 27 May 2015 at 22:57, Eugene Alvin Villar sea...@gmail.com wrote: A possible problem is that currently, Wikidata notability policy[1] means that Wikidata will only contain items for notable objects/entities/concepts. (But note that Wikidata is much, much more inclusive than Wikipedia—Wikidata will contain vastly more items than Wikipedia has articles.) This means that not all buildings, streets, and other objects that we have in OSM will have corresponding Wikidata items. Wikidata (and for that matter the English Wikipedia) notability policies allow for an item for every settlement; even a hamlet of just a few houses. The issues on naming described at the top of this thread are unlikely to apply to small objects, such as individual dwellings and minor roads. That said, Wikidata also allows for an entry for every designated historic monument (listed building), and already has an entry for every street in the Netherlands. -- Andy Mabbett @pigsonthewing http://pigsonthewing.org.uk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
2015-05-27 23:13 GMT+02:00 Frederik Ramm frede...@remote.org: The place node for London has 154 name tags as we speak, but there are several thousand languages in use on the planet, so there's still room for enhancement. wasn't our credo that what the mappers are interested in will be accepted? There may be several thousand languages in the world but it is not a given that all those will end up in the OSM db sooner or later. Not only well-known tourist magnets carry foreign names; some dedicated language mappers have gone over and beyond the call of duty and added, for example, name:ru tags even to small villages: It is difficult to judge when such foreign names have a right to be there, and when they're just inventions or name translations or transliterations. actually most real names are such, name translations or transliterations. To give a famous example that most people will know: The Colosseo or Anfiteatro di Flavio (someone decided for whatever reason that this was an old_name while formerly it was an alt_name) http://www.openstreetmap.org/relation/1834818 is called in English: Colosseum and in German Kolosseum or alternatively: en: Flavian Amphitheatre de: Flavisches Amphitheater Those are clearly translations and transliterations (K instead of C) but they are at the same time very established names in these languages. This is equally true for other famous monuments like the fori imperiali (de:Kaiserforen), mercati traiani (de:Trajansmärkte), Bocca della Verità (de: Mund der Wahrheit), Fontana di Trevi (de:Trevibrunnen), but there are also exceptions, e.g. the en:Spanish Steps / de:Spanische Treppe are called Scalinata di Trinità dei Monti in the local language (it is located at piazza di Spagna, that's where the foreign name comes from, while in Italian it is called after to church it leads to). Naturally, OSM has the original name of this world famous monument, but Wikidata hasn't. Wait, it hasn't the original name of this three-star-tourist-attraction, how's that? Have a look here: http://www.wikidata.org/wiki/Q848072 the reason is that the Italian wikipedia hasn't got an article about the steps, they are featured in the article about Piazza di Spagna: http://it.wikipedia.org/wiki/Piazza_di_Spagna If we were to massively use wikidata _instead of duplicating some details from there also in our db_ we would have to improve wikidata as well, and impose our entity structure on them, or it won't work in some cases (and if it doesn't work in some case, it doesn't work at all). Another issue I see with wikidata is that it contains information and details about spatial objects, but it doesn't contain the geometry it refers to. Have a look at the Berlin object: http://www.wikidata.org/wiki/Q64 This covers both, an administrative entity and a geographic place in one object (no problem here, but can be a problem elsewhere). This object has a property instance of metropolis http://www.wikidata.org/wiki/Q200250 I don't want to discuss whether Berlin is a metropolis or not, what I want to point out is that there seem to be different criteria defined for different languages: English http://www.wikidata.org/wiki/Q200250?setlang=en metropolis very large and significant city or urban area -- generic, has significance as absolute criterion, not related to a region or country, no details in which fields significance is required German http://www.wikidata.org/wiki/Q200250?setlang=de Metropole Großstadt, die den politischen, sozialen und wirtschaftlichen Mittelpunkt eines Landes bildet -- has to be the political, social and economic centre of a _country_ (relative definition) Italian http://www.wikidata.org/wiki/Q200250?setlang=it metropoli città di grandi dimensioni la cui area metropolitana raggiunge o supera i cinque milioni di abitanti -- big city whose metropolitan area is = 5 million inhabitants French http://www.wikidata.org/wiki/Q200250?setlang=fr Métropole No description defined The Italian definition speaks about a metropolitan area, but there is no reference (neither in OSM nor in wikidata) about the extension of this area. The same problem is generally there with population: we cannot see in wikidata to which area the population refers to (and OK, we cannot be 100% sure in OSM that the population really refers to the area that is drawn ;-) ). metropolis is a subclass of big city: http://www.wikidata.org/wiki/Q1549591 which defines all cities with more than 100.000 inhabitants as big city (a German definition, which I believe will not hold true for China or other very populated regions). Indeed in OSM we do not have this hard limit (any more): http://wiki.openstreetmap.org/wiki/Tag:place%3Dcity but rather use the more clever definition: The largest urban settlement or settlements within the territory. My conclusion: I'd rather prefer to keep names in different languages inside OSM, because it makes it clear to which object they refer, while it is less clear
Re: [OSM-talk] Can wikidata links help fight name inflation?
2015-05-28 11:58 GMT+02:00 Frederik Ramm frede...@remote.org: I think that OSM is a database of local knowledge and culture, not of remote knowledge and culture added from afar. Therefore I find it out of place for OSM to see that objects like the London node receive a constant flow of edits from people whose only link to London is that they happen to speak a language in which London has a different name. how did you verify this? I feel that there are two totally different planes of editing - one is what's on the ground in London, mapped by people in London, and the other is what name the Martian civilization has chosen to give to London, something in which Londoners have no say whatsoever. You are trying to disqualify foreign languages in London by saying they were Martian, which they surely aren't, I'll bet on this if you like ;-) Are you aware that in London there are more than 250 languages spoken? People living there, not just visitors. Here's a link: http://resources.woodlands-junior.kent.sch.uk/customs/questions/regions/languages.htm London is special in this context, as it is a truely global city like there are very few (NYC, Tokyo, Hongkong, maybe a few others like Paris, Dubai, Moskow (?), ...) The name:xx tags are, if you will, the only tags for which the local mappers are not, and cannot be, the ultimate authority. And that's what makes name:xx stick out like a sore thumb for me; in my mind, OSM is first and foremost a project that lets the people control their map, instead of being told and labelled from afar. please acknowledge that there are minorities in every bigger place that do actively speak those languages. They ARE local (often). Cheers, Martin ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On 28/05/2015 11:20, Iván Sánchez Ortega wrote: El Jueves 28. mayo 2015 10.59.21 Steve Doerr escribió: There might be a case for adding pronunciations (of 'difficult' names at least) to the OSM database. Someone must have proposed a tagging scheme for this, surely? Yup. https://wiki.openstreetmap.org/wiki/Proposed_features/Phonetics name:pronunciation, as mentioned on that page, is in use in a few problems, and would surely solve the Slough problem: http://www.openstreetmap.org/node/10021975/history (though John Betjeman's idea might have been better) Cheers, Andy ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
W dniu 28.05.2015 12:41, Mateusz Konieczny napisał(a): Further complicating such edits by moving it to Wikidata or somewhere else is in my opinion a bad idea. We would rather retrieve it from Wikidata, because many places are already there! Nova Scotia? - you're welcome: https://www.wikidata.org/wiki/Q1952 With a editor completion you could have much more not going anywhere outside: - international name variations (including polish) - direct link to a rich objects database (with a broad context relations - even outside the scope of a GIS if you need it) Speaking of relations - there's even the link to OSM object already: https://www.openstreetmap.org/relation/390558 So what's the problem? It is also a bad idea to add thousands of names without a really good source and verification. Especially automated adding name:xx based on transliteration alone is a terrible idea that should be reverted once spotted - sometimes there is a separate name in foreign language, with difference going beyond transliteration. The data in Wikimedia projects should be verifiable, so this would be just a double check. We can do it, sure, but why not to start with something rather than from zero? What may be done is to improve editor interface to do not display 100+ name:xx tags for places like London. It would be nice improvement indeed - Wikimedia services already hides most of interwiki links - but it has nothing to do with Wikidata as a helper for OSM objects names and categorization. -- The train is always on time / The trick is to be ready to put your bags down [A. Cohen] ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
On Wed, May 27, 2015 at 11:13:11PM +0200, Frederik Ramm wrote: we're seeing more and more name:xx tags on OSM objects. [...] Generally I don't think having names in different languages on OSM objects is wrong. We have done it that way for a long time and, lets face it, as long as we allow it for some objects and languages, people will add names to other objects and in other languages, too. Moving responsibility to a different database is nice in theory, but not in practice. The common mapper will not understand why *his* language should not appear, but only some other and he will not understand how to edit a different database (especially not wikidata with its horrible interface) unless we build something for him to make it easy and seamless. Not only well-known tourist magnets carry foreign names; some dedicated language mappers have gone over and beyond the call of duty and added, for example, name:ru tags even to small villages: http://taginfo.openstreetmap.org/keys/name%3Aru#map (This is a matter currently under investigation by Data Working Group and it is relatively certain that not all 582,653 name:ru tags will remain.) Thats a different matter. While there are many names for London in different languages, I don't think there are special Russian names for half a million places on Earth. Chances are they are the result of automatic transliteration. And results of automatic processes should not be mapped for obvious reasons. Jochen -- Jochen Topf joc...@remote.org http://www.jochentopf.com/ +49-351-31778688 ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
Hi, On 05/28/2015 10:50 AM, Martin Koppenhoefer wrote: The place node for London has 154 name tags as we speak, but there are several thousand languages in use on the planet, so there's still room for enhancement. wasn't our credo that what the mappers are interested in will be accepted? No, that would be stretching the definition of credo. I think there's a high level of agreement for stuff that's on the ground and not of a too temporary nature has a place in OSM. Stuff that's not on the ground - e.g. the name given to a place by a culture thousands of miles away - can occasionally be allowed but there are limits. actually most real names are such, name translations or transliterations. To give a famous example that most people will know: Yup, a speaker of English will come home from a Rome visit and say I was at the Colosseum. - They will not, to repeat my favourite example, come home from a visit to Paris and say I was at the New Bridge. But let's not get sidetracked, that's a different discussion from the Wikidata question. I just hope that Wikidata doesn't list New Brige as the English name of Pont Neuf or else they have a problem ;) If we were to massively use wikidata _instead of duplicating some details from there also in our db_ we would have to improve wikidata as well, It is my impression that a large proportion of name:xx tags in OSM are added by naming specialists who do little else than large scale name additions; it would probably not be too much to ask for them to indulge Wikidata instead of OSM. and impose our entity structure on them, or it won't work in some cases (and if it doesn't work in some case, it doesn't work at all). I don't agree. If we could offload 99.99% of all name:xx tags to Wikidata and keep them only in edge cases like your Scalinata di Trinità dei Monti, why not? Why would a few cases in which name:xx tags remain ruin the whole scheme? I don't want to discuss whether Berlin is a metropolis or not, what I want to point out is that there seem to be different criteria defined for different languages: You're listing some interesting shortcomings of Wikidata that I weren't aware of. However these would not, in my opinion, bar us from using Wikidata as a name repository for rendering; if a mapper is of the opinion that no matching Wikidata object exists for an OSM feature, then they shouldn't use a wikidata tag, that much is clear! I agree that tag lists of hundreds of names in different languages aren't very handy to look through, They're also very hard to verify; and verifiability is important for OSM. I've seen small villages in England which had a name and a name:ru tag, and the only occurrence of the name:ru tag on a web search was on a dubious Russian weather and events in random city page. When is a name a name? Bye Frederik -- Frederik Ramm ## eMail frede...@remote.org ## N49°00'09 E008°23'33 ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] Can wikidata links help fight name inflation?
W dniu 28.05.2015 10:50, Martin Koppenhoefer napisał(a): My conclusion: I'd rather prefer to keep names in different languages inside OSM, because it makes it clear to which object they refer, while it is less clear from wikidata. Also because the structure of osm and wikidata is not the same, it will lead to problems (either we'll be/risk making links that are not precisely 1:1 or we'll have to change the structure to meet (either in wikidata or in OSM)). Placenames are geographic information that do belong into OSM IMHO. I was sure that in every two (or more) projects there will be some differences and that's why I said about forking Wikidata - that way we would gain (wild guess) 95% of existing objects from the start and we would still be able to have local (delta or differential) version for remaining few percent. It's tempting to have control over everything we use, but the NIH ( https://en.wikipedia.org/wiki/Not_invented_here ) syndrome is counterproductive in the longer run. And while OSM is big enough now, it will be much bigger in next years (still many parts of the world outside the West are poorly mapped!) and using already mature, rich, free-licensed and community-controlled resources would likely help us with that. Every big project will have to deal with inconsistencies or ambiguities and there's no escape - not because we're careless, but the world is complicated. So better to prepare for real global mapping than to be afraid of the risks. It's too late! We're not just a London-centered street map anymore, so let's face the global class of problems. I agree that tag lists of hundreds of names in different languages aren't very handy to look through, but IMHO we should resolve this in the GUI (display name translations in the editors closed so you have to click on an arrow to unfold the list, or sth like this, and/or let the user set a list of languages he want't to see the names in and hide the rest under a single line like i18n names, 182 tags, etc.) I prefer GUI for linking the name of the proper Wikidata entry with the possibility of: - forking the object name into our diff database - merging local and Wikidata data if/once they're ready to do it. -- The train is always on time / The trick is to be ready to put your bags down [A. Cohen] ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk