Re: Facebook Linked Data

Am 26.09.2011 um 23:37 schrieb Patrick Logan:

> On Mon, Sep 26, 2011 at 1:53 PM, Sebastian Schaffert
> <sebastian.schaffert@salzburgresearch.at> wrote:
>> 
>> the point here is that this "philosophical" discussion...
> 
> I did not see anything philosophical in the discussion. I am confused
> by that statement.

The philosophical part is the distinction of

http://graph.facebook.com/561666514 representing the document
vs.
http://graph.facebook.com/561666514# representing the person. Which it, btw, does not, because I never gave Facebook the exclusive permission to provide the URI for me as a person. At best, it describes a user account.

The distinction is impractical, especially if you compare it with the simplicity of the alternative JSON-based Open Graph representation.

> 
>> When I request a Linked Data resource, I expect what comes back to describe what I requested. Otherwise, I as a developer have no way of knowing how I should access the data: when I request a resource that is a document and I get back a description of a person (i.e. other subject) how should I know where to start?
> 
> It looks to me as though you received exactly what Jesse described in
> his announcement...
> 
> "If you don't have a vanity URL but know your Facebook ID, you can use
> that instead (which is actually the fundamental method)."


No. 
- I ask for http://graph.facebook.com/sebastian.schaffert and I get http://graph.facebook.com/561666514#
- I ask for http://graph.facebook.com/561666514 and I get http://graph.facebook.com/561666514#
- I ask for http://graph.facebook.com/561666514# and I get 404.

I have no simple generic way of deciding how what I requested relates to what I get back. Of course I could do it specifically for the FB Linked Data service, but then it would not work for other LD services.


> 
>> Developers need well-defined behaviour, and not "this service does it like this, and that service does it like that". This is not the interoperability promise the Semantic Web gives.
> 
> Certainly greater levels of compatibility will take time across the
> wide-open internet. But this message seems to be saying, "I did not
> heed the description of the service as it was announced, and the
> result surprised me, therefore the service is not practical"???


Maybe I misunderstood the interoperability vision of Linked Data. But if I have to follow different service descriptions for different services, I can also stick with RESTful services and JSON. The whole point of Linked Data for me would be to be more independent from the different service descriptions and be able to follow a standardised, well-defined procedure for getting data.

Impractical is: if I write a Linked Data client application I have to follow different service descriptions for each Linked Data service I am accessing.


Sebastian
-- 
| Dr. Sebastian Schaffert          sebastian.schaffert@salzburgresearch.at
| Salzburg Research Forschungsgesellschaft  http://www.salzburgresearch.at
| Head of Knowledge and Media Technologies Group          +43 662 2288 423
| Jakob-Haringer Strasse 5/II
| A-5020 Salzburg

Received on Monday, 26 September 2011 22:09:30 UTC