[SWSL] What's distinctive about Semantic *Web _Services_* (was Re: today's minutes)

To www-wsers, a bit of context.

In the last SWSL (Semantic Web Services Initiative, Language Committee) 
telecon, I raise the question, What's distinctively *web* or *web 
services* about the language we're defining. At the moment,  looking at 
our list of objectives and requirments, and reflecting on our 
deliberations, I couldn't think of anything.

This made me wonder if we'd been remiss in not thinking more deeply 
about whether there *is* something special about the Web, Web Services, 
and the Semantic Web that should inform our work. Perhaps we're just at 
SUCH a high level of abstraction that nothing from below interestingly 
affects us. (If so, I think we shouldn't *be* the *Sementic WEB* etc. 
etc.)

Part of the problem is that both the Web and Web services attempt to 
be, in some sense, universal and embracing. The Web *engulfs* other 
information systems (sometimes, with the mere flash of an URI scheme 
registration). Web Services seem to do something similar (are there 
*any* restrictions on bindings? in principle?)

If we're just making convenient use of some current standards, but 
really couldn't care at all about any of their supposedly distinctive 
features, well, I'm not sure what we're doing then :)

Some replies inline.

On Sunday, October 12, 2003, at 02:05 PM, Terry R. Payne wrote:

[snip]
>> What I'm curious is, other than a broadening acceptance and
>>> understanding of wide-area service integration, and perhaps a bunch
> of
>>> useful standards (e.g. WSDL), what are Web Service technologies
>>> actually
>>> giving us?
>>
>> What of *Web* technologies themselves?
>
> Useful, most definitely, but I firmly believe we should accommodate the
> developments in WS standards, but not be constrained by them.

I'm only interested in WS standards, from a theoretical point of view, 
in so far as they embody something distinct and interesting, preferably 
Web like. I hear the term Service Oriented Archtiecture. I read debates 
between REST folks and SOA folks. There seem to be distinctions that 
make differences, and yet we don't seem sensitive to tehm.

>  For
> example, I recently had a discussion with someone who had made the
> assumption that all messages between WSDL clients would be strongly
> typed, and formatted in XML (which isn't always the case - consider 
> http
> bindings, for instance).

That would be being constrained by the accidental and the incidental. 
I'm curious to know if there's anything fundemental, perhaps 
architectural, that we can exploit or must respect. (Wow, I got a lot 
of -al words in there!)

>> (And it might be that the Web or Web Services constrain us in ways
> that
>> something like CORBA might not. One decision is whether to overcome
>> those constrains (e.g., by building up another layer) or to work
> within
>> them because we think it gets us something.)
>
> We should harness what WS give us, but... well I'm about to repeat what
> I said above.

What *do* they give us? What's accident and what's essence?

>  Also, I'm seriously concerned that we could risk
> re-inventing the wheel, by ignoring work in other service-anaolgous
> research fields (e.g. DAI, Grid, Multi-Agents etc).
[snip]

That's a related but different concern. We all think we're trying to 
give *added value* to WS, and thus, we hope, the Web. So of course, 
we're not *entirely* bound, but presumably we want to augment, not just 
layer on.

Cheers,
Bijan Parsia.

Received on Sunday, 12 October 2003 14:48:54 UTC