xml-dev - Re: Web service and Semantic Web

["W. E. Perry" <wperry@fiduciary.com>]

Whether or not I can answer N. Uramoto's question to his satisfaction, I am very
grateful for this succinct statement of it. IMHO, he has pinpointed precisely
where Semantic Web enthusiasm runs off the rails:  the question of how, and based
on what, the requisite inference is performed. In the past several days on this
list we have seen at least three threads raise similar questions of pre-ordained
semantics. Whether the question arises with regard to namespaces, or to schemas,
or to messaging specifications, the underlying problem is the same.

As Semantic Web advocates continue to elaborate their vision, it is increasingly
clear that they imagine a lattice of interlocking references, definitions and
assertions. Within such a scheme, resolving, or instantiating, or otherwise
realizing the particular manifestation of any particular node or component of
this lattice on any particular occasion necessarily involves inferences,
principally as to how to preserve and express the semantics of referenced
resources within this new realization. Because this Semantic Web rests on a
fundamental premise of building new assertions upon reference to previous ones,
it relies not only upon the fixity of those assertions but also upon a mechanism
which can be counted on to constrain any new expression to the scope, or more
precisely the intersecting scope, of the underlying referenced assertions on
which it is predicated. Autonomous nodes, implementing idiosyncratic methods,
cannot be counted on to respect that limitation.

The distinction, then, between 'Web Service' and 'Semantic Web' is whether the
nodes of that web can be trusted, by the promulgators of the larger scheme, to
implement unique (and perhaps uniquely useful) functionality. Stated another way,
the difference is whether the outcome or rendition from a node is governed
principally by its own functionality or principally by the received semantics on
which it operates. That distinction is so basic that it is not surprising that
Uramoto-san wonders 'What is the relationship between them? Are they completely
different things? or similar but supported by different communities?' The answer,
I fear is that proponents of both camps do inhabit this list, where far too often
they are simply talking past each other. The gulf may, in fact, be unbridgeable,
but now at least we have the specific differences laid out for any who might
relish the debate.

Respectfully,

Walter Perry


URAMOTO Naohiko wrote:

> Web Service:
> The Web is a set of software components (methods)
> No inference
> Ontology
> We can find services from open directories such as UDDI
> Digital signature is used for integrity and non-repudiation
>
> Semantic Web:
> The Web is a set of assertions
> Inference
> Ontology
> (How we can find assertions?)
> Digital signature is used for qualifying assertions (I think it is not correct
> use of digital signature...)