Re: Linked Data Browser

If everyone had a Linked Data browser, I wouldn't need to bother creating the html myself.

My objective might well be to publish RDF - there is no reason (for some of my datasets) to publish html at all, as there is no expectation that humans will want to read it.
So, if a human does, they can do the work of formatting it into a human-readable form, if they want.

I often start with just the RDF, for example in an RDF store, and serve RDF in whatever form I feel like, because that's what the service is for.
Solely creating turtle or json-ld or whatever, which is cheaper to produce (and cheaper for the consumer) than constructing HTML+RDFa seems to be a fine option.
It is as frictionless as possible.

I have nothing against RDFa - I think it is great, and should probably receive more attention around these lists, but using it for machine to machine communication is not ideal, or, I think, what it is optimised for.
For me, Linked Data and the Semantic Web are about machine-to-machine communication, and RDFa embedded in HTML sort of assumes that there are humans around at every stage (not just start and finish, perhaps), and the objective is for machines to be better able to interpret what the human-oriented document is saying.
That is not always the case.

It all depends whether you think the source format is RDF or HTML, I think.
I think I may be starting to repeat my self ;-)

best
Hugh

> On 16 Mar 2018, at 18:28, Sarven Capadisli <info@csarven.ca> wrote:
> 
> On 2018-03-15 19:49, Jean-Marc Vanel wrote:
>> You're right about RDFa ,
>> but since not everyone does serve RDFa, there is a need for a RDF browser.
> 
> A need for who or what exactly? Please describe the use case.
> 
> The website/service already does conneg for HTML (without any RDF
> embedded) and RDF/XML. It has some sort of a templating mechanism to at
> least output different representations (aside: at the moment the content
> is not equivalent in both of those representations). In any case, this
> is a matter of updating the existing HTML template to include RDFa
> alongside existing human visible content.
> 
> 6 RDF browsers were originally linked. None are currently functional. Do
> you think that a yet another round of RDF browsers are going to resolve
> this issue, indefinitely?
> 
> Right now the following URLs are floating out there:
> 
> http://km.aifb.kit.edu/projects/numbers/n7
> http://km.aifb.kit.edu/projects/numbers/web/n7
> http://km.aifb.kit.edu/projects/numbers/data/n7
> 
> There is virtually no added benefit to managing all of those URI spaces,
> where one would reach the same goal. That's a far more complex system
> design than it needs to be, both to publish and to consume.
> 
>> Moreover, Linked_Data_Browser allows one to browse on the real RDF URI's
>> , whereas the RDFa pages are more an artifact on the view side (there
>> can be no relation between URI's exported by a RDFa page and the page URL).
> 
> Define "real".
> 
> Whatever that can be represented in Turtle can be in HTML+RDFa. Provide
> a simple Turtle example and I'll show the equivalent in RDFa, and you
> can get the isomorphic graph back, and then in Turtle.
> 
>> Also, I wanted to stress that RDFa is *not the only way* to offer a
>> human readable view of RDF data . HTML + RDF + Turtle, etc, *content
>> negotiation* is another way , more in the line of Tim's original view of
>> the Semantic Web. For instance , the URL that I mentioned in this thread,
>> http://semantic-forms.cc:9112/ldp/semantic_forms
>> <http://semantic-forms.cc:9112/ldp/semantic_forms>
>> can provide Turtle , RDF/XML, JSON-LD or HTML , depending on HTTP
>> "Accept" header.
>> dbPedia and Wikidata also do this .
> Conneg is orthogonal. No one suggested RDFa in HTML is the only way. My
> whole point was about describing a simple(st) system that can serve for
> humans and machines, which happens to be possible via single URL. No out
> of band requirements to discover or manipulate the data.
> 
> Turtle or JSON-LD data islands in HTML tend to duplicate information,
> and are hidden from human view. Hence, they require additional
> processing, like relying on JavaScript or a supplemental server
> algorithm to build a "human view" (HTML) eventually. Not to mention that
> the processor instructions are not as clear for JSON-LD/Turtle in HTML
> as it is for RDFa. There is also the rabbit-hole of handling bnodes and
> namespaces across multiple HTML script blocks.
> 
> So, why should the server bother to output a Turtle or JSON-LD dump if
> it can be equally said to output RDFa in HTML? There are use cases to
> include Turtle or JSON-LD, but I just don't think that's the case for
> this particular website.
> 
> -Sarven
> http://csarven.ca/#i
> 

-- 
Hugh
023 8061 5652

Received on Friday, 16 March 2018 19:59:14 UTC