xml-dev - RE: RDF, the "semantic web", and the nadir of AI (was RE: Realist icprop

["Bullard, Claude L (Len)" <clbullar@ingr.com>]

<warning>If you are reading this when you should be coding, 
you are guilty of defrauding your company.  Shut up and 
go back to work.</warning>

I don't disagree with your sketch of the interactive web. 
But it is interactive now.  Semantic?  I don't think so. 
Could use more annotations or meta information?  Sure.

Systems that gather information on the actions of the user 
and feed these back into the system to say, generate a user 
interface or alter information such as navigation, are tactical, 
the so called, Type C production systems.  These are also the 
basis for the Class 4 and 5 IETM systems that use pre and 
post conditions to choose nodes and diagnose faults.  
This is where the web is already at depending on just how hard 
a site developer wants to work at it.  For example, the 
pornography industry works very hard at it and shares information 
extensively.  DoubleClick works.  Whatever your tastes are, 
they are busily trying to get new and different content to 
satisfy them.  Caveat emptor.

Steve Newcomb has a good comment on such systems:  "Don't 
put anything on the web that that you wouldn't want to 
see on the front page of the New York Times or have 
your minister read to the congregation in the Sunday 
morning service."  You must understand policy and apply 
it well.  For that reason, the governments of the world 
are involved and will increase their involvement in the 
web business.  You get to trust elected officials or 
company officials or hackers.  Consider the subtext of the 
new credit card commercials in which people shopping online 
are dropped into a basement hell made up of unshaven stinky  
young programmers taunting the shopper about how they are smart 
enough to steal and the shopper isn't smart enough to stop them.  
The shopper holds up the magic card, and poof, all the stinky 
people vanish to replaced by the shining white light of 
professional commerce.

You can see the zeitgeist forming. Yes, it is a stereotype, 
and yes, it is sticking.

By clickstream mechanics, form-filling, full-text analysis, 
annotation and so forth, a great deal of information can 
be aggregated and inferences made.  Inferences backed up 
with confidence weights etc are all part of the AI legacy 
that we understand and yes, are applying to web 
systems.  No problem.  A coarse transaction that can be 
stored and restored enables this to work just fine.   We 
can do all of that. Yet another old tech becomes new again. 
That's good.

I've seen semantic network systems, neural networks, 
rule-based forward and backward chaining for expert 
systems, and the whole bit that was AI before it went 
out of vogue.   If someone creates services based on these 
and improves capabilities, they may improve the products. 
That the entire future of the web will be based on these 
is doubtful.  That the future of the successful business 
is and will continue to be reliable services with high quality 
is not in doubt.  If RDF, topic maps, and XLinks can 
increase this reliability and quality, then they will 
play a role in that future.  Put all of that in the 
context of an n-tier architecture (presentation layer, 
workflow layer, business objects, database) and we 
can easily talk requirements and get results.

As you say, "once we have the data, we can do anything 
with it".  On the other hand, others will do unto you 
as well with their data.  The realism of the proposals 
to the W3C must be their basis in promoting 
not just advanced, but demonstrably useful services.  
That may be what TimBL means by "does socially useful 
things".  It is meaningful because it means something 
to the user.  The word is not the thing, as the linguists tell us.

Be sure your policies promote trust.  The essence is 
stable cooperating systems, alliances, not complexity 
as a barrier to competition.   Discoverable services 
must have clear, simple, reliable interfaces.

Len 
http://www.mp3.com/LenBullard

Ekam sat.h, Vipraah bahudhaa vadanti.
Daamyata. Datta. Dayadhvam.h