RE: Has ISSUE-66 been properly addressed?

> -----Original Message-----
> From: Erik Wilde [mailto:dret@berkeley.edu]
> Sent: Friday, August 08, 2014 4:20 PM
> To: Markus Lanthaler; public-hydra@w3.org
> Subject: Re: Has ISSUE-66 been properly addressed?
> 
> hello markus.
> 
> On 2014-08-08, 2:43 , Markus Lanthaler wrote:
> > ISSUE-66 [1] was raised by Erik. There have been lots of discussions
> > and as far as I can tell, Ruben has implemented all feedback in the
> > spec apart my comment in [2]. I don't see that as a blocker though.
> > Has this issue been adequately addressed?
> > I'm primarily asking you, Erik, here (CCed) as you raised this issue
> > in the first place but of course everyone is welcome to comment.
> 
> i am fine with what was proposed (thanks, ruben, for your suggestions), and
> apologize for the traffic volume. i was trying (unsuccessfully) to avoid the
> well-known perma-discussion, accusations, and conspiracy theories.
> 
> my only point was to raise the issue that i meet a lot of people who are not
> aware of the fact that linked data has a narrow definition, and just
> understand the term as "data that is linked". explaining this is, imho, the price
> you have to pay when you use very natural-language term to mean
> something specific.
> 

Erik, I would have thought that simply pointing those people to TimBL's 5-stars of Linked Data [1], and pointing out that the stars represent a progression, and that you only have 'Linked Data' when you achieve all 5-stars (as David Booth explains nicely in [2]), would be a very easy and cheap 'price to pay'.

Sure, they'll ask 'what are RDF and SPARQL?', but a simple answer to that is 'well, they're only important if you really need interoperable data between parties that don't know each other' or 'they're for helping with data integration when you have many disparate systems', or even 'well, you don't need to worry about them yet, just use JSON-LD for now and we'll come back to them later if needed!'. (But if those people really *do* need to integrate across very many disparate datasets today, then I'd strongly suggest they read up on RDF straight away, even just for education on 'prior-art'!)

So I just read the entire mail thread point to by Luca [3], and frankly I now agree with David Booth even more strongly than I did originally. I think any disconnect may be because most people today just aren't that interested in genuinely interoperable data. They're still happy to provide point-to-point data integrations that require manual mappings between their datasets (i.e. 'no-one will take vitamins now if it will only help them in 5 years time!' [4]).

I think lots of those people simply aren't aware of the possibilities opened up by RDF (hence they need to ask 'what are RDF and SPARQL?', like lots of people in my organisation), while many of those that are aware of them just don't care (because they are only looking to integrate, or even just expose to 'the Web', very few datasets). Therefore this very sizeable percentage of the industry don't want to be bothered with learning new stuff like RDF and SPARQL, because they genuinely think they won't ever need them (and lots of them will actually be right).

I think it ties in very well with the comment from Markus about the current batch of Hypermedia API approaches [5]:

"Most approaches treat the data as an opaque blob, i.e., they are not really interested in it. Instead, they focus on adding mechanisms to express links and in some cases also operations to JSON."

I strongly agree with this appraisal, and personally I think it's kinda short-sighted of those approaches - at least from my professional perspective, as I have lots of disparate datasets to integrate. The other approaches may be fine for one-off's, or quick wins, or small scale integration, but I just don't feel they fit my current use-cases. And so that's why I personally favor Hydra (and the fact that it's underpinned by RDF) right now.

But if Hydra and LDF can help to bridge the gap between these two perspectives of 'Linked Data' (and I think the huge success of JSON-LD at least suggests that this may very well be possible), then I genuinely think that everyone can actually win here.

In the meantime, I'll certainly be sticking to the formal RDF-based basis for *my* definition of Linked Data... :) !

[1] http://www.w3.org/DesignIssues/LinkedData.html 
[2] http://lists.w3.org/Archives/Public/public-lod/2013Jun/0396.html 
[3] http://lists.w3.org/Archives/Public/public-lod/2013Jun/thread.html#msg208 
[4] Paraphrase of quote from the Hypermedia API discussion recently!
[5] http://lists.w3.org/Archives/Public/public-hydra/2014Aug/0064.html



> over and out,
> 
> dret.

Received on Sunday, 10 August 2014 12:16:15 UTC