Re: SEMIS Bulletin
Enrico,
You used the word "hardly" several times, but without
being specific. Please elaborate.
JS> Except for the prettier graphics and simpler links
> of the WWW, I was doing the same kinds of things I do
> today on IBM's VNET in the 1970s and 1980s.
EF> Hardly.
IBM's VNET in the 1970s and 80s supported email, instant
messaging, file transer, shared files, bulletin boards,
news groups, remote printing, remote log-on, free download
of IBM software and documentation, a wide range of remote
services, and the ability to create your own remote server
just by writing a program that would respond to messages.
In fact, the VNET protocols were much easier to support by
writing a simple program than the current web protocols.
I used to write my own email handler because it was very
easy to implement with just a few lines of code.
When I visited any IBM location anywhere in the world
(includng Europe, Japan, Argentina, Brazil, and Egypt),
I could log on to my userid at Yorktown and do everything
I could at home, although with some delay. It didn't support
http, but I could browse a wide range of files, services,
and discussion groups. I could also link to Arpanet for
email, ftp, and other services.
I don't understand what you mean by "hardly".
JS> I agree with Pete that EXPRESS comes much closer to the
> goals of "well-defined meaning" than RDF and OWL.
EF> Hardly. STEP and EXPRESS are very interesting and successful
> representation languages, which have many advantages (the most
> important being that they are really used in practice), but
> which still miss a serious and complete "well-defined meaning"
> in terms of logic: formal semantics, definition of reasoning
> problems grounded on the semantics, complexity of reasoning
> problems, correct/complete inference algorithms for the reasoning
> problems, complexity and well founded optimisations for the
> inference algorithms). RDF and OWL (well, not OWL-full) do
> already have all this, and the attempt anyway is to go in this
> direction.
I certainly agree that EXPRESS should be based on the Common
Logic foundation. But it already has the expressive power
of first-order logic, and some people have already written
translators to and from EXPRESS and versions of FOL, such as
KIF. I don't disagree with you that all those theoretical
niceties are very nice. But they pale in significance in
the light of three very simple facts:
1. 99.9% of the people who use RDF, OWL, Express, SQL, or
any other kn. rep. notation don't have a clue about
complexity, completeness, well-foundedness, and other
theoretical issues.
2. The methods they use for debugging their representations
are the same methods they use for software design and
development: trial and error.
3. EXPRESS supports a richer modeling language than OWL,
which is essential for representing the full range of
structures and operations required for manufacturing.
I agree with you that good logic-based tools are highly
desirable, especially for those people who don't know logic.
But you can certainly develop such tools for Express or SQL
as well as for any other kn. rep. language.
It seems obvious that EXPRESS is a vast improvement over OWL,
and again, I fail to understand what you mean by "hardly".
JS> In summary, my recommendation is to replace the term
> "semantic web" with "semantic system" and to integrate all
> related technologies -- including SQL, UML, EXPRESS, OWL,
> RDF, and others -- on a common logic-based foundation.
EF> Look, this is already happening in the semantic web:
> it is an open world, and many of us are exactly working
> in this direction (e.g., [1] for a survey)
I'm glad that you are doing that, so am I and so are a lot of
other people. But what bothers me is that the W3C started at
such a low level with RDF that most of what's been written
so far is, from a logical point of view, garbage. Compared
to what's been done in RDF, the work in SQL and Express, as
untheoretical as it may be, is better structured. I seriously
doubt that most people will use OWL in a more logical way.
JS> The human interface should be controlled NLs supplemented
> with graphics.
EF> Again, this is being explored as well since ages.
That's what I said -- but not only "explored", it's been
implemented for ages. In 1971, Fred and Bozena Thompson at
Cal Tech had a nice implementation of REL (Rapidly Extensible
Language), which was a version of controlled English that was
as easy or easier to adapt to new applications than XML.
EF> XML is only a convenient machine-readable syntax. What do
> you have against it? It is content neutral - unlike RDF.
As I said, XML is a great notation for document markup (it's a
variant of IBM's GML, which I had been using for over 30 years).
But compared to LISP or dozens of other systems (such as REL),
it is a clumsy, inefficient basis for defining other languages.
John