Changeset: 63797799
name:et
Closed by Pikse
Tags
created_by | JOSM/1.5 (10966 et) |
---|---|
source | http://www.eki.ee/knab/knab.htm |
Discussion
-
Comment from Nakaner
Hi,
you have added Estonian names for various locations all over the world from a third party source in this changeset.
When did you discuss this import on the Imports mailing list as required by the Import Guidelines? https://wiki.openstreetmap.org/wiki/Import/Guidelines
Where is the explicit permission of the copyright owner documented which allows us to use the data? The license statement at the source you linked is not compatible with our licensing:
"The data of KNAB may be freely used provided that the source is quoted. Extensive usage of the data (e.g. if a monograph, gazetteer or a map is based on the data of KNAB) will be possible on the presumption that the compiler of the database be informed of this usage beforehand. There will be no charge for the use of data, instead the user might receive information on the completeness and reliability of the data he/she needs."
Please pause your import until these issues are resolved.
Best regards
Michael
-
Comment from Pikse
This is the source that I've used for copying/checking individual names. I didn't consider this as an import. I've queried nodes based on place/population key value and then I've tried to provide adequate localized names. In order to make sure that names are reliable I've checked names from this database or from print atlas (referenced in another changeset).
I'm not sure what extent of use of this database amounts to "extensive usage". I've verified a few hundreds of names based on it, which is a tiny portion of about 478K names that the database has. I didn't copy any additional name variants for individual places nor other metadata specific to this database. So I'd rather think I haven't infringed rights of the maker of database.
For the most part these names are romanized based on rules that are not exclusive to this database nor is any general idea on how to romanize names copyrightable. So I could romanize these names myself based on the same rules. This would be just considerable more time consuming for me. And if I did then it'd be still favourable to check the result against source like this database.
Are you sure you are not overly rigid about this? Or what would be more preferable way to localize place names then? After all, individual names are not copyrighted and it should be possible to localize every single place name without acquiring any permission.
-
Comment from Nakaner
Hi,
if it is not clear how to understand the terms of use, OSM usually takes a whiter than white view and refrains from using the data.
The terms of use describe a gazetter as "extensive usage". OSM allows any use of its data, a gazetter is quite common (e.g. Nominatim). The terms require us to informed the compiler of the database before any use. Our license does not require that. It does not require a credit to KNAB.
The license is not compatible with OSM from my point of view. Could you please revert your edits (remove added tags, change modified tags back, …)?
You are free to disagree with me. I appreciate it if you use the Talk or Legal-talk mailing list in that case.
Best regards
Michael
-
Comment from Pikse
Terms of use that you refer to do not really tell that using, say, ten names or a few hundred names in a product like gazetter is "extensive use". I discussed this aspect above you haven't explained where in this matter lies the reasonable doubt that given use might be extensive.
I've read import guidelines and I now also read some other copyright related OSM documentation. To sum up my previous comment I don't see why do you think that import guidelines apply here in the first place. This in my opinion is not aligned with common sense.
Should someone really copy significant amount of data from KNAB, say, including coordinates for place name placement, specific selection of name variants for objects etc. that'd allow recognizing KNAB as data source then sure, I'd understand why you'd think that sui generis database rights or other sorts of rights of the maker of database might be infringed. In this case however use of this database amounts to merely checking what is the common place name in Estonian for individual places.
Another way to go would be to not provide any source at all, and say that name variants given are common/preferred in Estonian. So that anyone knowing the language and subject can compare the data addition with any reliable source that they like (or re-romanize them using appropriate rules). However, I believe it's more friendly towards other OSM contributors wanting to review to provide some particular direction for review (as I did here).
Note that there can be two distinct purposes for providing a source. One is verifiability and another is attributing the author. My intended use for source key here was the first one. To my understanding use of source key is not restricted to attribution.
I've also checked histories for various place name label nodes that I've touched. For name variants added in other languages I generally don't see reference to any source or permission. Source would be nice for these, but making into a legal problem probably wouldn't be reasonable as an individual name variant used in a language usually really is a common knowledge.
So, I don't think that I've really violated a policy or some other reasonable and common practise here, and I don't see a good reason for revert. As explained above, currently it's unclear what the actual legal problem might here, and I don't know what exactly should be discussed in a mailing list. If you still think that there is reasonable doubt about infringing someone's rights, then I'd be grateful if you did bring this up in an appropriate venue, so that my comments given above were reflected and so that it would be referenced here.
-
Comment from Nakaner
Hi Pikse,
I contacted the OSMF Data Working Group and would like to ask you to pause the "import" until they respond.
Best regards
Michael
-
Comment from Pikse
I haven't added any localized names after your first comment here. I also wasn't planning to add more names at the moment since I'm more or less done with providing Estonian names for major cities (based on place/population key value). However I expect that work on adding localized names should continue in future and I still expect that in order to ensure data quality some sort of source should be provided for verification.
For further background information, my work on these names relates to Wikimedia's effort to provide localized map tiles for its projects. Compared to larger languages number of names available for Estonian is rather low at the moment. There's a Wikimedia blog post about it [1].
Regarding data working group's input on this, am I going to be contacted about this as a related party? Or should I make my own enquiry? So that I'd know what exactly did you ask, and in order to know how to provide a source for verification so that it wasn't mistaken for an import source.
[1] https://blog.wikimedia.org/2018/06/28/interactive-maps-now-in-your-language/
-
Comment from Nakaner
If you come from the Wikimedia community, please read https://wiki.openstreetmap.org/wiki/Welcome_to_Wikipedia_users It describes the main differences between Wikipedia/Wikidata and OSM.
Please also note that name:<lang>=* should not be added if it is a transliteration only which can be generated by a computer programme automatically. That is done already for the map at openstreetmap.de which transliterates all non-Latin names if no German, English or French name is tagged. (It's open source)
The map service of WMF does not transliterate automatically but that's a problem of their service and must not be a reason to add names to OSM. Such issues have to be resolve by the consumer. https://wiki.openstreetmap.org/wiki/Tagging_for_the_renderer
The Data Working Group will comment to this changeset. You don't have to do anything yourself.
-
Comment from Pikse
I've been around for a while I generally understand how OSM works. Though, I must admit that this "no transliteration" section on "Names" wiki page is new to me. I must say it seems to be written by someone who doesn't know different romanization systems quite well. Different languages use different systems, there are many exceptions and so insuficent results are pretty much inevitable. Moreover for languages like Arabic meaningful transliteration is pretty much impossible as vowels are generally omitted in Arabic. You can check youself on openstreetmap.de that it can't really handle place names adequately in e.g. Saudia Arabia unless tag with a Latin-script name is present. So using automatic transliteration in my opinion is a very bad idea.
I'm not sure how widely supported is this "avoid tranliteration" approach. After all, as shown by statistics referred in blog post referenced above, then providing localized names on OSM is actually very common. Though, I must admit of course that should majority of OSM users really advocates for "avoid tranliteration" then this is a problem for Wikimedia's current approach.
Nonetheless, I don't think that "Tagging for the renderer" is appropriate judgement here. After all, common name variant used in particular language is not specific to any renderer, it's specific to language, and I don't see why any localized rendered wouldn't be interested in consuming this data (or at least prefer it any more or less worthless automatic "tranliteration").
-
Comment from Pikse
Now that I look further, wiki talk page [1] clarifies that three users were chatting that they didn't like what a particular user/tool did (or they didn't understand what it did) and then this chat was documented as "avoid transliteration". Doesn't look like a real policy to me.
[1] https://wiki.openstreetmap.org/wiki/Talk:Names#Avoid_transliteration.2C_why.3F
-
Comment from Pikse
To further clarify how I've processed place name translations: for large part of these I copied the key value for English-language name and then checked if it's the same in Estonian. This is especially the case for Chinese names that usually are the same.
If there are any further questions about my work or the specification of KNAB database then I'm happy to answer.
-
Comment from woodpeck
Hello Pikse, I also think that your approach is questionable. Apart from potential legal issues with the data source, I have a hard time seeing the value of a name:es tag being added to a small city in Asia with the same content already present in name:en (or even in the main name tag). Nakaner is right about our approach regarding transliterations; if you want to question this please do it in a larger forum (e.g. talk mailing list) instead of simply ignoring it because "it doesn't look like a policy to me". There's also a general unsolved problem with the name:xx tags and that is that we in OSM usually require that things should be verifiable on the ground. If someone claims "the name of this place is Blah", and I say "I don't believe you", then we can always go to the place and ask the locals. Now if you take the name for some small city in Asia from the Insitute for the Estonian Language and add it - who's going to verify that? We'll have a hard time finding an Estonian who lives there and who says "yes, this is really what we Estonians here call the place". So I would at least recommend a very cautious approach when adding foreign-language names to places.
-
Comment from Pikse
I get that OSM values that its contributors map things in nature as they are and hence the "verifiable on the ground" principle. However, OSM also includes features that cannot be mapped in nature and that definitely cannot be verified on the ground, for example administrative boundaries that have to be verified by other means. The same way it'd make very little sense to me if "verifiable on the ground" was applied to place names. Current guidelines seem to suggest that "name" key and it's variants provide name variant that is common in given language (local language for "name"). Name variant used by local residents of given place (village, town etc.) is not necessarily the same. If it isn't the same then it's a very specific kind of place name that you generally don't see on maps. Verifying common name in given language amounts to knowing conventions in given language, verifying local residents' name variant amounts to survey as part of ethnographic research (this does not mean "ask random resident"). Neither variant is really verifiable on the ground.
Argument that you make against using name:es for small city in Africa is really an argument against providing name variants in any language, including name:en, and even main "name" key as well as for many places there simply is no local resident that you can ask (especially the case for names of natural features). I understand it's up to OSM community to decide that they are not interested any data that isn't verifiable on the ground, just like it has decided it's not interested in e.g. elevation data. Currently however this does not seem to be the case as features like administrative boundaries and place names are common in OSM.
There is another page that describes why translating names is useful. My work on that generally aligns to it:
https://wiki.openstreetmap.org/wiki/Map_internationalizationYou probably know there are world atlases in all kinds of languages. These atlases usually follow conventions of given language to write place names. For example, French atlas wouldn't omit place name in Africa merely because no Frenchman happens to live there, nor would it write it in non-Latin script.
I'm having hard time understanding how it'd possible to provide adequate place name data without doing it language by language and place by place. As already pointed out above, different languages use different romanization systems, and there are many exceptions to use of particular system in particular source and/or target language. Any fallbacks to English or automatic transliteration inevitably leads to a mess.About avoiding transliteration I said "it doesn't look like a policy to me" only because referred page clarifies that there apparently was no wider discussion prior to writing it down. On same page you can see that it already has been questioned a while ago for obvious reasons. I'm planning to comment there as well later.
Database by the Institute for the Estonian Language is simply a prominent source in Estonian language. It's made by people who most likely know the subject and who are aware of conventions in onomastics in Estonian. It gives an overview of place names that you'd generally find on published/edited Estonian-language maps. This database is rather meant to be used for verification rather than meant to be verified itself.
About potential legal issue, in addition to what has been said already. Given kind of use is similar to using a dictionary. For example, if I checked many words from dictionary when helping to write some guideline for OSM or when participating in a discussion like this, then you would accuse me of violating copyrights of those who put together the dictionary or the database behind it?
- أبها (295924875), v13
- الدمام (287865393), v14
- الديوانية (248421092), v21
- Agadir ⴰⴳⴰⴷⵉⵔ أگادیر (27565078), v66
- 阿克苏市 (244075609), v29
- البيضاء (288491858), v21
- الباحة (295921642), v15
- الفلوجة (192581952), v20
- الفيوم (768555545), v17
- الفاشر (300074306), v13
- الإسماعيلية (768595872), v20
- الجهراء (1278188834), v15
- الكوت (599473544), v15
- المنصورة (31582434), v24
- المحلة الكبرى (31582451), v14
- المكلا (5579116461), v4
- المسيب (2093558057), v8
- القضارف (294865520), v24
- القرنة (2487904414), v9
- الأبيض (248948689), v17
Welcome to OpenStreetMap!
OpenStreetMap is a map of the world, created by people like you and free to use under an open license.
Hosting is supported by Fastly, OSMF corporate members, and other partners.
https://openstreetmap.org/copyright | https://openstreetmap.org |
Copyright OpenStreetMap and contributors, under an open license |