Copyright applies to actual text (or other forms of expression), not to ideas or the content. Since the items do not allow to enter wikitext, there is no compatibility issue due to the license. Filling the language links based on the existing language links in the Wikipedias is not a problem re copyright. Cheers, Denny 2012/11/15 Daniel Kinzler <daniel.kinzler(a)wikimedia.de>de>: -- Project director Wikidata Wikimedia Deutschland e.V. | Obentrautstr. 72 | 10963 Berlin Tel. +49-30-219 158 26-0 | http://wikimedia.de Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e.V. Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg unter der Nummer 23855 B. Als gemeinnützig anerkannt durch das Finanzamt für Körperschaften I Berlin, Steuernummer 27/681/51985.

I agree. Just to clarify what "actual text" should mean: Although a short sentence with several words may occasionally be a copyrightable text (e.g. a poem), it is very rarely so. On Wikipedia infoboxes, due to scope, purpose and style, this can almost be excluded. It is not desirable to exclude brief scope notes or source notes, which occasionally occur in Wikipedia infoboxes, just because they contain several words. I personally would recommend an extraction dryrun and manually check for parameters that have more than perhaps 12-15 words, whether they are creative (= copyrightable) or plain expressions of fact or sources (= not copyrightable). Gregor

On 15 November 2012 23:35, Sven <svenmanguard(a)gmail.com> wrote: (The argument above is independent of licensing, so this should perhaps be discussed in a separate thread?) Gregor

Maybe I've missed something because I'm coming late to this, but it surprises me that people do not seem concerned about keeping track of licenses (not just credit) for the data we'll host. At least in the EU, recently compiled or modified databases are protected by database rights. Maybe the WMF is not affected by such restrictions, and only needs to respect U.S. law. But many Wikidata editors do live in the EU, and so do many potential re-users of the data, in other Wikimedia projects and beyond. I'm concerned that failing to keep track of data licensing will put them at unnecessary legal risk, and will hinder Wikimedia's wider mission of providing free educational material. For instance, once Wikidata content is incorporated into Wikipedia articles, projects like Wikipedia CD Selection ( http://en.wikipedia.org/wiki/Wikipedia:Wikipedia_CD_Selection) might not be able to freely release their DVD of selected Wikipedia articles in the EU unless it respects the licenses under which the data has been released. (Perhaps this is the case already, but Wikidata will make it much more visible to data owners.) We will find it very difficult to respect these licenses if we don't keep track of them from the beginning. Recording the source of the data is a vital part of this, but it isn't sufficient. For instance, U.K. government data is often released under the Open Government License ( http://www.nationalarchives.gov.uk/doc/open-government-licence/), which requires not just credit but (where possible) a link to the license itself. What to do with data from the EU that's not freely licensed is a thornier issue. Just adding it to Wikidata is arguably illegal and against WMF's terms of use, at least for people living in the EU. I'm aware that many people feel database rights are a bad idea (including me). But they are part of the law in many countries, including Hungary (where phase 2 will be launched), so we shouldn't just ignore them. Going back to the original question, I think the database of interwiki links in Wikipedia is unlikely to creative enough to be copyrightable. The CC-BY-SA-3.0-unported license used on Wikipedia is silent about database rights, so databases that are not copyrighted do not trigger the BY-SA requirements of the license. So I believe Wikidata's current work on interwiki links doesn't violate CC-BY-SA. Avenue

Just to clarify, my concern is about externally made databases, regardless of whether these are imported directly into Wikidata, or have been incorporated into Wikipedia first and imported into Wikidata from there. For example, the population data in Wikipedia's list of ceremonial English counties ( http://en.wikipedia.org/wiki/List_of_ceremonial_counties_of_England), which also features in the infoboxes of the articles on each county, would I think be covered by database right under U.K law. Like other ONS material, it has been made available under the OGL, which does impose some obligations on re-users (somewhat similar to CC-BY). I'm not so worried about databases originally generated within Wikipedia, e.g. interwiki links. I think I agree with Gregor's analysis of these, if I understand him correctly. I'm no lawyer either. I'd be very happy to be told I'm wrong, that we have had legal advice and there's no need to worry about licensing external data. Avenue

I guess things would be far more easier if WikiData would choose a CCBYSA-compatible database license like ODbL. I can only repeat myself, that the OpenStreetMap community went through a similar discussion and decided to respect database rights and keep their data as clean as possible. Because they didn't want to take the legal risk, that after 5 years a big company like Google will sue them and so OSM might loose the work of 5 years. I also think that there are two aspects in this topic: 1) the pure legal aspects and 2) how we (Wikipedia users) treat each other and respect the work of others. So the WMF might say: hey you European guys with your DB rights... bad luck that you invest all your work into a US hosted project - we don't care about your rights. Even _if_ this is legally not to be objected, is it fair against the users? To make it clear: I'm into lots of list articles (streets and monuments) which contains lots of data. This content is licenced under CCBYSA today. The day when all this content will be expropriated by WikiData, is the day I'll stop my work for these Wikipedia projects. Regards Alex -- http://de.wikipedia.org/wiki/Benutzer:Alexrk2

No you cannot: CC-BY-SA is not intended for data, so regarding data you should never use any CC licence (pre-version 4.0, v4 will address database rights) other than CC0. That's why if you want BY-SA conditions, the good choice is ODbL which was created by the Open Knowledge Foundation exactly for this, a share alike licence for opendata. Like I already said a few months back, I totally agree with Alex : using ODbL and therefore keep along the BY-SA choice that rules Wikipedia is the most logical choice. Mocing towards CC0 would be a source of a lot of possible conflicts and possible infringements of Wikipedia's content licence. Benjamin Ooghe-Tabanou from Regards Citoyens (France) On Wed, Nov 28, 2012 at 5:23 PM, Denny Vrandečić <denny.vrandecic(a)wikimedia.de> wrote:

I think the situation is different here, the unported CC-BY-SA doesn't grant any database rights, so it might not be possible to use the data with CC0 or ODBL (it is debatable, that's the point where consulting with a lawyer is useful). http://wiki.creativecommons.org/Data has some information. On Wed, Nov 28, 2012 at 5:43 PM, Denny Vrandečić < denny.vrandecic(a)wikimedia.de&gt; wrote:

On 28/11/12 17:58, Luca Martinelli wrote: Dual licensing under ODBL + CC-BY-SA? Or even ODBL + CC-BY-SA + GFDL, to keep the legacy license as well. You could also ODBL with an additional permission to relicense it under CC-BY-SA or GFDL when aggregated with another work (slightly different than a dual licensing, but pretty much the same).

There is some confusion in this thread. A single data item is not licensable. The fact that Berlin is the capital of Germany, or that the population of the Seychelles is 84,000 is not licensable. Let us hope that this will never change. Copyright covers a specific expression, e.g. a concrete text, image, etc. For example, the Britannica article on Germany or the Wikipedia article on the Seychelles might be copyrighted, and thus can be licensed with a license like CC-BY-SA as is the case for the Wikipedia article. This is why it makes sense to allow for different licenses on Commons, at it has many different images and each might have a different license. In a few jurisdictions there are sui generis database rights. These cover complete databases - but they do not cover their individual data items. There are licenses built on top of these database rights, and ODBL is the one that is used by OSM. Unfortunately, CC-BY-SA and ODBL are not compatible. You can not take content licensed under one license and republish it under the other license. Thus follows that using ODBL for Wikidata does not relieve us from *any* of the possible legal issues that are mentioned here due to Wikidata using CC0 as a license. Having different licenses for different data items in Wikidata, as was suggested here, is not possible, as single data items are not licensable. For every data item in Wikidata, there will be the possibility to add a reference that supports that data item. The license of that reference has no effect on the license of the data item. This is the same that happens in Wikipedia: in the article on Japan we might and do have references that have proprietary copyright. We might have a reference that has a CC-BY-ND-NC license. But that does not mean that the article on Japan has to be CC-BY-ND-NC. The license of the reference has no effect on the license of the Wikipedia content. If Wikidata was a collection of databases, like e.g. OKFN's DataHub, then it would make sense to provide for different licenses for each of these databases. But Wikidata does not have any notion of different databases. You cannot take a database and simply upload it to Wikidata. You can do that on DataHub. Wikidata is not DataHub, and does not aim to be DataHub. DataHub is pretty awesome at being DataHub. The goal of Wikidata is different. Choosing CC0 now allows us to later switch to CC-BY-SA4 if we choose so, which actually does take care of database rights, unlike previous versions. Choosing ODBL would not allow us to do so. I hope that clarifies the discussion a bit. It is a murky legal area, which has not been much tested yet in courts, so there are some insecurities there. I hope this helps, Denny 2012/11/30 Finn Aarup Nielsen &lt;fn(a)imm.dtu.dk&gt; -- Project director Wikidata Wikimedia Deutschland e.V. | Obentrautstr. 72 | 10963 Berlin Tel. +49-30-219 158 26-0 | http://wikimedia.de Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e.V. Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg unter der Nummer 23855 B. Als gemeinnützig anerkannt durch das Finanzamt für Körperschaften I Berlin, Steuernummer 27/681/51985.

2012/11/30 Avenue &lt;avenue42(a)gmail.com&gt; Wikidata can not and will not host whole databases. I said that explicitly int he last Email and I am repeating it here. Wikidata is not a collection of databases. That is also not supported by our software. For this use cases, CKAN is the appropriate software and DataHub would be an appropriate public instance of CKAN. You can not upload a database to Wikidata. This is no feature that we intend to implement. Wikidata can not host whole databases. Wikidata can not and will not host different databases. This is the central issue. Wikidata is *not* a collection of databases. The notion of different databases does not exist anywhere in Wikidata. It is not represented in the backend, it is not in the UI, it is nowhere to be found. No. For none of the promises of Wikidata is the upload of databases as a whole necessary. I hope this clears up some of the raised concerns. Cheers, Denny -- Project director Wikidata Wikimedia Deutschland e.V. | Obentrautstr. 72 | 10963 Berlin Tel. +49-30-219 158 26-0 | http://wikimedia.de Wikimedia Deutschland - Gesellschaft zur Förderung Freien Wissens e.V. Eingetragen im Vereinsregister des Amtsgerichts Berlin-Charlottenburg unter der Nummer 23855 B. Als gemeinnützig anerkannt durch das Finanzamt für Körperschaften I Berlin, Steuernummer 27/681/51985.

On 30/11/12 19:18, Sven Manguard wrote: I know the last mail concerning this is some days ago, but I just wanted to mention my understanding. Maybe it's helpful for "unweirding" the fact about database licenses. Compared to a database a text is the selection and arrangement of different words. The text with it's meaning or even no meaning (Lorem ipsum?) is copyrightable but not the single word. So I assume that single facts (or database items) are not copyrightable just like single words. Only the database (or even a view?) as a selection and arrangement of various items is copyrightable. Authors often have the problem, that there is already a text and they are not allowed to just copy this into Wikipedia. They rewrite it with their own collection of words, which can – of course – also reuse words from the original text, because those are not copyrightable. Thus it should also be possible to use the own collection of data-items that are also part of proprietary data bases to create an own, free-licensable database. I understand Wikibase as a kind of NOSQL-Database, with a list (it is not a real table) of items with internationalized titles and aliases. Each of this items can contain various properties, but it does not have to. So item1 can contain property1 which item2 does not contain. Therefore speaking of a table is IMHO not really correct, because the only real columns you have are ID, titles and aliases. While persons have a birthday, companies and organizations have a founding date. Cheers Marco

On Mon, Dec 10, 2012 at 10:10 PM, Gregor Hagedorn &lt;g.m.hagedorn(a)gmail.com&gt; wrote: While some databases are copyrightable, I expect this wouldn't usually apply to the sort of information we want to store. I believe we're more interested in collecting comprehensive sets of facts than creative selections and arrangements of them. The copyright also might not apply to the atomistic storage of those facts in Wikidata. The arrangement and presentation of those facts to readers in a Wikipedia article might more easily violate the copyright, although fair use might provide a defence there. Some individual statements might arguably be sufficiently creative in themselves to attract copyright, e.g. sharemarket indices. I'm not aware of any clear precedents for this, however; this is the closest I've seen: http://wjlta.wordpress.com/2010/07/15/federal-court-copyright-law-might-pro… So I don't see a great need to worry about copyright in Wikidata. If it does apply in some exceptional cases, we can deal with that when it arises. Database rights have a much wider effect, which I still think should be of concern. But that topic deserves an email of its own. Avenue

On Wed, Dec 12, 2012 at 3:22 AM, Tom Morris &lt;tfmorris(a)gmail.com&gt; wrote: a) None. I'd be very happy to hear from people who do. b) The US, at least. I did say that IANAL more than once earlier in this thread. Maybe I should make it part of my signature. Avenue

Maybe I'm missing something, but why should we restrict Wikidata to a single standard license (or dual-license)? Why not be like Commons, where content can be uploaded under various free licenses? This would have the advantage of letting us respect the licensing schemes of various data sources. For example, much UK govt data is provided under the OGL, while other organisations release data under a ODBL or a CC-BY license. For user-collated data, we could still encourage or even require a certain license (or dual license). Avenue

On Thu, Nov 29, 2012 at 9:42 PM, Alexrk I was thinking that licenses would apply to a dataset, not to individual items. I don't believe database rights apply to individual items, only to entire datasets (or substantial portions of them), so it would make sense that data licenses would apply at the same level. I'd imagine that we don't want to store databases composed of items that require individual licenses (e.g. images or lengthy texts), at least to begin with. The status of "produced works" would presumably depend on the provisions of the data license. Some might be placed under the same license, or require attribution (e.g. for the ODBL), or not require a license at all. Deciding it's all too hard may be one option. But we should be clearly aware of the long-term consequences of not tracking licenses before we make that decision. I'm not convinced we have thought this through enough yet. Avenue

On Thu, Nov 29, 2012 at 11:06 PM, Platonides &lt;platonides(a)gmail.com&gt; wrote: Yes, but in some jurisdictions the licenses are the only things allowing lawful use of the data (at least in sizeable portions). I'm concerned that we'd put some users, and maybe entire projects, at unnecessary legal risk if we ignore this. If the graph in your example was based on 100 numbers from one database, and 100 from another database, then the licenses for both source databases could apply to the derived database. If the licenses weren't compatible, that would indeed be problematic in countries that recognise database rights. To me, this seems yet another reason to keep track of the licenses, so we don't unintentionally lead people into legal trouble. On the other hand, combining data is not always problematic. If each population number in your example was from a separate database, extracting a single number from each database should not infringe on their database rights, and so their various licenses would not apply. (Of course, we'd still want to cite the various sources.) The resulting collection should be a fresh database that could then be hosted under any free license. Avenue