sameAs.org license - was Re: Linked data sets for evaluating interlinking?

Thanks Ghislain.

Sorry, no SPARQL endpoint, as it isn't an RDF store.

With respect to a license, it is more difficult.
This may be a longer answer than you were expecting. :-)
(Firstly, please understand that I'm not very good with this license stuff.)
When I started sameAs.org, it only had mostly my rkb stuff in it.
So I could do what I liked.
Understanding the importance of having some sort of license, I put what I thought was the most liberal one I could find - http://creativecommons.org/publicdomain/zero/1.0/ (which is at the bottom of the page).
"Take it away and do what you like with it".
I would have liked to say "Please attribute if you can, but I understand that may be difficult, so don't worry if you can't" but I couldn't find one of this, and I think having a license that is quickly seen and widely understood is important.

<Sub-bit on attribution>
A problem with follow your nose (fyn) Linked Data is that the attribution can be very hard.
I may tell you that <a> owl:sameAs <b>.
The reason I tell you that is that I have found loads of stuff about <c>, <d> and <e> which allowed me to infer that. And some of that data may no longer even be available.
So the only safe attribution for every "fact" I give you would be my entire source attribution - I might as well tell you the attribution is the Web.
Correct, but hardly in the spirit of the thing (I am actually more interested in the spirit of fair attribution tun the legal side of it!)
For one of my users who uses fyn to attribute is probably even harder - at least I now my sources by hand.
If they came using fyn, then they may be using a URI that happened to be got by a previous resolution (and so on).
So essentially, every time they resolve a URI, they need to do license work.
Of course in principle this is what people should be doing - absolutely!
But in practice, people are not tooled up for this; so a requirement for attribution would make the data unusable for such people.
And they are the ones who are *really* using Linked Data, so I want to encourage them!
</Sub-bit on attribution>

Of course, it now has stuff from lots of other sources.
Many of these simply sent me the data, or told me I could put it in sameAs.org - but I don't really recall anyone ever discussing license!
Since I asked for it for sameAs.org, then I assumed that they agreed to have it out there with the license.

Other stuff, I have just gone to a sparql endpoint or download site and taken a bit of their data.
So what is the license of stuff on the open web? - No, you don't need to answer that!
Essentially sameAs.org is a search engine for the Linked Data web; so I went to Google and Bing to see what license they might put on their data.
Answer found I none!
[I even found that if you put a search such as "Bing license" into Bing it barfs! :-) ]
There is lots of stuff about what users license them to do with user data, and what they license for their software, but nothing on the results returned from a web search on their site.
My sameAs.org "about" page does list a bunch of places which should provide compliance with any attribution requirements for those sites, but is now seriously out of date, I think.
So I just left it at that.
I know I don't have the same legal department as Google or Microsoft if there is a problem :-), but I sort of think that I take far less data from sites than they do, and it doesn't seem to be a problem for them.

As far as the sub-stores are concerned, I took the license off.
But most of them were built in collaboration with the sources, and they have links to the sources, which may or may not have a license, but that probably makes things clearer for those.

The bottom line is that there are very few sites, if any, which (like sameAs.org) have as their main purpose the provision of sameAs information.
On the contrary (like googlejuice SEO) they want any sameAs links to be taken away, so that traffic will come to their sites through the links they have published (like via Google).

Thanks for your question - I'm happy to get any advice from anyone, and I hope I can understand it if it comes!

Best
Hugh

On 27 Aug 2013, at 09:09, Ghislain Atemezing <auguste.atemezing@eurecom.fr>
 wrote:

> Hi Hugh,
> 
> 
>> So, for example, if you wanted Adrian's data, then I can give it to you.
>> (I have queried the SPARQL endpoint to put stuff in sameAs.org. Both owl:sameAs and skos:exactMatch.)
>> I have lots of bibliographic ones, especially national libraries, who have often sent me the data.
>> (British, German, US, Japanese, Norwegian, French, Spanish, Hungarian … as best I recall.)
>> I also have the VIAF data.
>> This is all aggregated in http://sameas.org/store/kelle/ and other stuff is kept in some sameAs stores - see http://sameas.org/store/
> 
> Nice work!! And a small question…..
> I was wondering if there is an endpoint in sameAs.org for using SPARQL queries? 
> And for the data sets you receive, do they all have a specific terms of license? which one/ones?
> 
> Thanks again for maintaining such a great service!
> 
> Cheers,
> Ghislain
>

Received on Tuesday, 27 August 2013 10:51:31 UTC