Two SemWebs (Re: GRDDL and OWL/XML)

I'm trying to sort through all the points raised (maybe this is too  
much). There are two poles of evaluation for me, technical and social/ 
marketing. I think both are important.

I rate this point low on both sides.

On 13 May 2008, at 19:07, Booth, David (HP Software - Boston) wrote:

>
>> From: Bijan Parsia
>> [ . . . ]
>> In the OWL/XML case we have:
>>         1) an existing specification (at the W3C) for a format  
>> closely
>> identified with semweb
>
> It seems to me that the lack of an executable GRDDL transformation  
> for OWL/XML would push us toward two separate Semantic Webs: The  
> RDF-based Semantic Web and the OWL/XML-based Semantic Web.

Given the availability of converters, given the prior existence of  
variant formats that require custom parsers even for the same model  
(e.g., Turtle), I would say that whatever push there is is at best  
nominal and the dangers highly, highly speculative. Furthermore, if  
such a problem does arise, and it is significant, there are many ways  
to route around it.

> IMO, there must be only one Semantic Web,

This doesn't seem to be a technical argument. I'm not sure what the  
advantages or disadvantages are, or what "one" (vs. "two") even  
*means* here. The regular Web survives with a wide variety of formats  
and with translation to a common model (i.e., html and the DOM) being  
accomplished in client and third party software.

It seems strange to argue that the Semantic Web is, as a matter of  
architecture, necessarily less robust than the Web. Indeed, this is  
an antimarketing point.

> and its lingua franca must be RDF.

This is an anti-marketing point, esp. when dealing with people who  
are not RDF centric. In the case of me and OWL/XML, I feel like we  
*have* paid more than lip service to RDF as an exchange format, e.g.,  
by developing the mapping and providing translation software. In the  
general case, RDF is held up with *such scorn* that forcing people to  
do something which involves a lot of strange antipatterns (spec by  
XML, automatic plugins from one provider, etc.) that are *not* the  
norm on the web just seems to feed that scorn (and, I would say,  
rightfully so). I thought that the point of GRDDL, from a *social*  
perspective, is to minimize the resentment and, well, laughter  
occuring when people suggest RDF as the formalism for some new web  
format. I seem to remember that you were in the WSDL group with me :)  
but I see we have drawn different lessons.

>   The value of providing an executable GRDDL transformation from  
> OWL/XML to RDF far outweighs the risk that it may conflict with the  
> OWL2 spec.

Mere assertion. Please provide SOME evidence. Right now I don't see  
any benefit to having executable GRDDL at all. I.e., what's the  
marginal gain?

>   Furthermore, if a conflict with the spec is found, the GRDDL  
> transformation can be fixed.

This is only one point, fwiw, but note that the lines of  
responsibility are quite dilute in this case. The WG will disperse.

> I think it would be wrong to assume that all tools that may benefit  
> from consuming OWL/XML will be sophisticated enough to know all  
> about OWL2

Why do you assume that all tools that may benefit from consuming OWL/ 
XML will have GRDDL support?

Why do you think it's very hard to support well known formats in  
one's GRDDL tools explicitly?

> or know where to find the best OWL/XML to RDF translator.

So you believe the *best* one will be what we provide at the  
namespace document? Isn't it quite the opposite? (As many people have  
conceded.) So now the burden isn't just a *correct* one but the  
*best* one?

Plus, uhm, I wonder how people manage with, y'know, every other piece  
of software out there. It seems that finding parsers and loaders etc.  
is NOT so very difficult.

> The reason for GRDDL in the first place is to avoid the problem of  
> requiring document consumers to have specialized prior knowledge of  
> every XML format that they wish to consume as RDF.  OWL/XML is such  
> a format also.

There is a big difference between having to know *every* and having  
to know *some*, well known, ones. This was my point about more ad hoc  
specs vs. W3C central ones.

Actually, I see no technical case here at all, and I see strong anti- 
marketing points.

Cheers,
Bijan.

Received on Monday, 19 May 2008 13:08:54 UTC