On Feb 1, 2009, at 11:08 PM, Mills Davis wrote:
Of course you've spoken in the spirit of criticism. No big deal.
No, really. I also admire John's work: he is a valued colleague and an old friend, after all. And I am in awe of the technology being produced by VivoMind. But I also have been closely involved with the development of the Semantic Web standards, and know first-hand how many different, often conflicting, perspectives and agendas had to be accommodated. Even if John were right about the technology (he isn't), there is no way that such a rigidly uncompromising position could have produced an acceptable standard. And whatever its problems, the fact is that RDF and OWL are being used by thousands of people and projects, and have become the standard way to communicate machine-readable information on the Web. Which was the point. Even if everyone translated OWL into Prolog before processing it (in fact, very few do), OWL/RDF would still be a successful standard: it was never intended to be a notation primarily for processing
information, but for communicating
it across the Web.
Personally, I think it is admirable that Sowa and others are addressing language understanding, a broader spectrum of reasoning, and techniques to make these scale logarithmically.
The semantic web is an important line of development, but its current goals are hardly the last word.
Well, its goals
are hard to fault, but I would agree that its current technologies are not the last word. Indeed, almost everyone involved with the SWeb believes this, which is why we expect that (like the non-semantic Web) it will continue to evolve. Still, for all their faults, the responsible attitude, it seems to me, is to set out to improve them, rather than simply take every opportunity to publicly attack or denigrate them passionately, especially when the criticisms are often rather shallow and uninformed.
I would add that when on the inside of the standards process, one often finds rather heated criticism from folk on the outside who feel that they have the one true solution, which has been unjustly or carelessly ignored. But the trouble is, of course, that there are many such one true solutions, and none of them are compatible with the others.
The sweet spot for semantic web technology today is linked data and infrastructure plumbing. For these opportunities there is usually more than one computing approach can be taken. The burden of proof is on semantic web advocates to show its relative business value -- which can be done, depending on application characteristics.
The real driver for semantic solutions are applications where the ontologies are dynamic, rather than relatively fixed. IARPA's Blackbook 2 for intelligence analysts comes to mind. Some dynamic infrastructure, information, process, and/or decision-making applications can be solved with semantic web technologies as we currently see them. However, semantic processing at scale remains a challenge for current semantic web approaches. More to the point, categories of problem exist that require knowledge representation that is beyond description logic. Similarly, there are types and combinations of reasoning needed that similarly are beyond the ken of current semantic web technologies.
Agreed. And I think nobody would disagree, even the authors of the OWL and OWL 2 specs. But what follows? Should we say, because other things will be needed one day, that nothing should be done now? With that thinking, no specifications would ever be written for anything. The most salient fact about OWL is that for a very large number of potential SWeb applications, it is too
expressive (and too complicated).
Its easy (and correct) to say that everything is just some subset of FOL. (Guha and I said it in the L-Base proposal, and Richard Fikes and Deb McGuinness had said essentially the same thing a few years earlier.) The fact remains that application developers need to have these subsets identified for them and packaged in some familiar and easy-to-understand-and-use form. And this isn't easy to do in a clean, general, way. We tried to do it with the RDF idea of a 'semantic extension', but that is too arcane and not sharply defined enough. The CL idea of a 'dialect' is related, but more to do with alternative syntaxes than logical subsets. Although its ugly, having distinct languages (like RDFS and OWL-DL and OWL-2-DL) for useful subsets of logic seems to be the currently best compromise, until someone can think of something better. And in the meantime, the world will create some neat applications using this scruffy, makeshift, shambolic compromise.
Let's embrace the semantic web for what it can do, and keep an open mind for more powerful and enabling technologies that will be needed as we move forward to the next internet.
Those technologies will be part of the semantic web. Its not a fixed set of technologies, but an ongoing, evolving, world-wide experiment. John is fond of trashing it in comparison with RDB technology; but RDBs have been around for 20 years, and they didn't create a semantic web.
BTW, thanks for the reference to MindCite.
On Feb 1, 2009, at 10:43 PM, Pat Hayes wrote:
On Feb 1, 2009, at 1:55 PM, John F. Sowa wrote:
... his standard, and by now getting rather repetitive, passionate and
almost comically biassed diatribe against RDF and OWL and XML and
anything to do with the Semantic Web. Of course, his small company is
free to adopt the operational strategy which suits them best. From
what I know of VivoMind, their core business does not appear to be
closely related to the goals of the SWeb. However, other small (and
some large) companies appear to disagree with him about RDF . As one
example, take a look at MindCite.
I point this out not in a spirit of criticism but only to provide some
reassurance to those who might find SW technology of interest. As on
a number of other topics, John's views are both idiosyncratic and very
starkly stated, admitting of very little discussion or debate. I am
not particularly fond of OWL myself, but to attribute all uses of it
to pointy-haired bosses who have drunk the kool-aid is misleading, and
indeed something of an insult to a large number of very smart people
of my acquaintance.
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
IHMC (850)434 8903 or (650)494 3973
40 South Alcaniz St. (850)202 4416 office
Pensacola (850)202 4440 fax
FL 32502 (850)291 0667 mobile