Re: SEMIS Bulletin
John,
DA> To put it another way, the WWW is the
DA> Semantic Web 80% done.
JS> My first objection is to the idea that the WWW
JS> makes a significant step toward the semantic web.
I can't speak from direct experience with Semantic Web
technologies, but I do think the Web makes a big
difference between earlier knowledge representation
efforts and Semantic Web initiative. A big one is
global scalability, where the key element is URI's
(and their already deployed global acceptance).
Hypertext was already a relatively mature technology in
the research community, when Tim Berners-Lee
hit upon what is needed to make it scalable: global
identification, a simplified stateless FTP that allows
intermediaries, and some simple representation loosely
based on SGML. I think the current efforts to "Webize"
knowledge representation (and at least DL approaches)
are worthwhile, and I wish the CG community would
participate more and address global scalability.
Another dimension that I think is important is
pragmatics, and I think the Web context of machine-
to-machine interaction (including semantic web
services) provides a source of new insights.
However I don't think the Semantic Web is anywhere
near 80% done, and that kind of presentation is a
disservice to the real accomplishments so far. Even
in terms of TimBL's own architectural stack, I can't see
anything like 80%. John's 8% is perhaps closer.
Surely though the greatest value of the Semantic
Web efforts is enlarging the Knowledge Representation
community, and attracting a lot of new talent. That
takes some "marketing" glitz, but I don't think the
proponents are particularly excessive. If anything,
I think they are underpromising, and are more
upfront about the difficulties than some of the expert
systems and knowledge-based systems advocates
in the early and mid eighties. And surely they have
been a great help in what I take to be your own
agenda, John, of balancing the inference side of logic
with detailed attention to conceptual schemes and
formal ontologies.
JS> I regard XML as a great notation for marking up
JS> documents. But it's more of a nuisance than a help
JS> for doing semantics. Instead of the 30-year-old
JS> GML base, I would prefer the 40-year-old LISP base
JS> as the foundation for a one-size-fits-all parser.
LISP has had its chance, and despite the seeding
of Scheme and LISP in many of the world's best
computer science students it is nowhere near XML's
global deployment. I think the pragmatic virtues of
XML as a self-labeling recursive data format go
well beyond representing documents, and a very
wide community of technology and content practitioners
seem to agree.
JS> And in any case, notation is the least of our
JS> problems when we're trying to do semantics.
In fact, I agree that notation is not itself that important, but
the Web's use of XML as a representation is more than just
a notational issue. Representation systems, and not just
XML but other data and metadata formats as well, are
architecturally crucial, especially when combined with URI's
and interaction (as the Representational State Transfer (REST)
camp insists to the more RPC-oriented web services advocates
who would close the world of representation and logic.)
And I don't think XML notation is stopping anybody.
N3, RELAX NG and XQuery all have moved forward
on human-friendly notations, while still retaining XML
interoperability.
I can't understand what issues are at stake in putting down
the semantic web. Is there some unspoken context over
there in the West, like the KBS vs. Neural Networks
competition for resources in earlier decades? Or are there
more substantive issues at stake? I think there are many on
these lists who are quite sympathetic to John Sowa's
perspective, and would like to have a better understanding
of what you think is wrong with the Semantic Web. In
particular, what would you do differently? If DL were
dumped for CG's would it make any difference?
(By the way, I bought your Knowledge Representation
book recently and it's great.)
Cheers,
Fred
Frederick B. Kintanar
Cebu City
John F. Sowa wrote:
> Danny,
>
> Your one-line summary reminds me of two of
> my other complaints about the semantic web:
>
> DA> To put it another way, the WWW is the
> > Semantic Web 80% done.
>
> My first objection is to the idea that the WWW
> makes a significant step toward the semantic web.
> The semantic part is vastly more difficult than
> designing a protocol for hooking together the
> Internet with a subset of SGML. I would put the
> WWW contribution closer to 8% or perhaps 0.8%.
>
> I regard XML as a great notation for marking up
> documents. But it's more of a nuisance than a help
> for doing semantics. Instead of the 30-year-old
> GML base, I would prefer the 40-year-old LISP base
> as the foundation for a one-size-fits-all parser.
>
> And in any case, notation is the least of our
> problems when we're trying to do semantics.
> (I say this as someone who has been doing a lot
> of work on notation, but I am the first to say
> that what is expressed is far more important
> than the notation for expressing it.)
>
> My second objection is that the word "web" in the
> phrase "semantic web" is the tail wagging the dog.
> I certainly agree that the network is just as
> fundamental to computing as the database, file
> system, utilities, applications, etc., etc., etc.
> And for that reason, it should not be distinguished
> as something separate.
>
> The title of the bulletin that started this thread
> is "Semantic Web and Information Systems" (SEMIS).
> I suggest that we delete the three middle words and
> address the problem of designing semantic systems.
>
> John