Thanks for the article. As you say, there seem to be a few teething troubles, but this should become a fantastic data resource once these are fixed and kudos to the NYT for making its data available in this way.

Is better tool support the way forward for this? Certainly more education should help, but there will always be pitfalls in representing information accurately. Many of the bugs you have pointed out should be identifiable automatically.

Anyway, I’m looking forward to being able to make use of the NYT data.

Cheers, Bill

(The link got mangled, I fixed it.)

A typically thorough treatment of the subject :) One thing you didn’t mention are links to resources that can provide some background on the licensing topic. As you know, Leigh Dodds (Talis), Jordan Hatcher (Open Data Commons), Kaitlin Thaney (Science Commons) and I ran a tutorial at ISWC2009 that addressed many of these issues. The slides from the tutorial should provide a useful introduction to people new to the subject:

http://iswc2009.semanticweb.org/wiki/index.php/ISWC_2009_Tutorials/Legal_and_Social_Frameworks_for_Sharing_Data_on_the_Web

The following paper from LDOW2008 also provides a more narrative context:

http://events.linkeddata.org/ldow2008/papers/08-miller-styles-open-data-commons.pdf

It is highly likely that we will work in the coming months to provide a more polished and comprehensive guide to licensing and waivers for Linked Data publishing.

Cheers, Tom.

(IMO, putting a URI into a literal in RDF is always a horrible idea, because it will confuse people to no end, and breeds interoperability nightmares.)

One point:

I agree that the values of the properties nyt:topicPage and cc:licence should be URIs instead of literals. But, apart from the convenience to receive clickable links by this change, my main reason is that using URIs instead of literals here makes it clearer that not the URIs but the resources denoted by these URIs are the things the triples talk about, namely the actual topic page and the licence file, respectively.

On the other hand, I would say that the value of the cc:attributeName property (or cc:Attribute) should really be a literal rather than a URI, since in this case it is the URI itself the property refers to, AFAIU. (But I think that a clever data browser will still be able to recognize such a URI literal as a URI and can then render it in a clickable way as well.)

Cheers, Michael

And thank you for your patience. It’s only been 4 months, since we announced that we committed to this path and we’re still learning all the particulars.

All the best,

Evan Sandhaus

Semantic Technologist New York Times R+D