[
Lists Home |
Date Index |
Thread Index
]
Mike Champion wrote:
> 11/8/2002 5:56:20 PM, Paul Prescod <paul@prescod.net> wrote:
>
>
>
>>It is quite possible that the Semantic Web technologies will be found to
>>have some fatal flaw, but it won't because the inventors neglected to
>>consider obvious questions like these.
>
>
> The Web is *full* of what were once considered
> "fatal flaws", but it just keeps going and going and going. I'm
> feeling like the semantic web is waiting and waiting and waiting.
> TimBL's been beating the drum for a long time now ... where's the parade?
When I first heard about structured markup in the early 90s, people were
already saying it wouldn't go anywhere because they had been hearing
about it for years and it hadn't materialized: "Sounds Great Maybe
Later." Later arrived.
When I first heard about Python in around 1997, it was already around 8
years old and yet hardly anyone had heard of it. Today, the level of
Python awareness out there is much higher and still growing rapidly.
Other technologies that took decades to achieve "overnight success"
include object oriented programming, Unix, hypertext and the Internet.
So I don't see any need to put a Best Before date on technologies.
> I know the questions are being asked, but are the answers helping people
> to do real work more productively than they could without SW technologies?
Smart people claim yes. Uche, Cowan, Borden etc.
> FWIW, every time I think that the time has come for me to take RDF, OWL,
> etc. more seriously and I look into its use for some specific problem in
> my life, I come away underwhelmed. I solicited testimonials
> from people who have had the opposite experience ... and have gotten mainly
> plugs from SW stakeholders.
I don't see that at all. For instance, I've gotten plugs from Uche
Ogbuji. He may be a stakeholder now, by virtue of spending effort
implementing RDF, but he did so in order to solve problems for his
consulting customers. As I understand it, he needed something that could
be queried like a relational database but was much more flexible in the
fast of changing schemas and complicated relationships. That seems like
a common enough problem and I can see how raw XML would not be much help.
> The only real "aha!" experience I've had on xml-dev when
> this permathread reappears was once when Jonathan Borden
> explained how a controlled vocabulary such as SNOMED in which the
> cross-cutting hierarchical relationships among medical terms can
> be put to productive use in querying and analyzing data.
> http://lists.xml.org/archives/xml-dev/200204/msg00944.html
> (which was inspired by http://www.openhealth.org/talks/XMLBioInformatics.ppt)
>
> If the kind of thing Dr. Borden is talking about is the 'semantic web',
> I'm hip!
Of course that's the semantic web! Do you think Tim B-L would disavow it? ;)
I think you're building a false wall between the semantic web and uses
of the semantic web technologies. It is a reminiscent of the problem of
AI researchers. Whenever something works it is no longer called AI
anymore so no matter what useful technology comes out of AI labs, AI
seems like a string of unmitigated failure.
> .. (although I don't work a field that has spent the last couple
> of centuries getting its vocabulary straight, sigh).
It seems to me that the flexibility of the semantic web technologies are
most important precisely when you _haven't_ got your vocabulary
straight. And I can't imagine that any medical doctor that medical
vocabulary is fixed and unchanging. Consider this quote from Jonathon
Borden in the thread:
"Think of DAML+OIL / WebOnt as a language for encoding SNOMED.
Particularly a language that allows the encoding to be put on the web
and one that allows bits of ontologies written by different groups using
different tools to be integrated together -- the tools _already_ emit
DAML+OIL."
If the ontology problem is solved and standardized, then why are people
still working on it? Also, according to the SNOMED website, SNOMED is
changing (mostly growing) quickly:
http://www.snomed.org/milestones_txt.html
And by the way, what do you think about this part of Jonathon Borden's talk:
Putting it all together
=======================
Biomedical information has many vocabularies - each in its own namespace
genetics “Bio ML”
pathology “SNOMED”
surgery “CPT”
medicine “ICD”
radiology “DICOM”
His solution: "DAML across schemas"
In other words he suggests to integrate a variety of controlled
vocabularies into a web of integrated information: a Semantic Web.
But if it's
> stuff like this (quoted from TimBL et al's SciAm article):
>
> "At the doctor's office, Lucy instructed her Semantic Web agent
> through her handheld Web browser. The agent promptly retrieved
> information about Mom's prescribed treatment from the doctor's
> agent, looked up several lists of providers, and checked for
> the ones in-plan for Mom's insurance within a 20-mile radius
> of her home and with a rating of excellent or very good on
> trusted rating services. It then began trying to find a match
> between available appointment times (supplied by the agents of
> individual providers through their Web sites) and Pete's and
> Lucy's busy schedules. "
There is no integration technology invented in the last forty years that
has not described scenarios like this as evidence of what can be done
once the technology is deployed. I guess XML is a pipe dream also,
because basically the same scenario is deployed to describe the value of
XML in _another_ Scientific American article:
http://www.sciam.com/article.cfm?articleID=0008C786-91DB-1CD6-B4A8809EC588EEDF
I'm sure I could find equivalent scenarios for Jini, TCP/IP, SOAP or
anything else.
There is nothing in that scenario that is technically impossible using
1970s technology. Or if we allow ourselves the luxury of 1990s
technology, we could easily sketch out the CORBA interfaces that would
make each of those queries possible.
What makes it _financially_ impossible is the price of integration. Each
new layer of technology reduces the price of integration which brings
the scenario closer to possibility. The SemWeb technologies are not
magical. They are simply the next logical step in the stack of data
integration technologies we have been building.
HTML is like text/plain except that it is easier for the computer to
interpret.
Domain-specific XML is like HTML except that it is easier for the
computer to interpet.
RDF+OWL is like XML except that it is easier for the computer to interpet.
Paul Prescod
|