xml-dev - RE: Soft Landing

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

Number One Advice:  Don't trust your life to a 
real time control system for at least three generations 
of deployment.  See Airbus disasters.  Most expert 
systems are toys but there are some very big and 
successful ones used in limited domains, for example, 
GE used these for mechanical control systems.  It 
was discovered that unless one *severely* limited 
the domain and precisely specified the types of 
services the system would provide, they were unreliable. 
In any case, they were very expensive.

Some URLs to look at for those who want to study semantic 
network theories and applications.

http://pespmc1.vub.ac.be/Papers/StructuringKnowledge.html
http://pespmc1.vub.ac.be/SEMNET.html
http://www.nlm.nih.gov/pubs/factsheets/umlssemn.html
http://www.nlm.nih.gov/research/umls/META3.HTML

Note use of terms such as reliable, coherent, transparent, and stable 
to describe models.  

Note: Personal interpretation plays a role for users of such models.  
For real time, mission critical decision making, such models are considered 
only in the role of a support service.  They do not enable the kind of
automated 
business services that Tim Berners-Lee describes.  Expert systems 
typically function as advisory services.  In a framework such as 
BizTalk where routing of processes to service application is a 
fundamental concept, these advisors may be used in cases where 
input/output processing has a low confidence factor.  The advisory 
service can offer help.  For example:

o Classification or epistemology - eg, according to this authoritative 
definition, this IS-A that

o Episodic inference - this set of inputs matches a historical event 
recorded in this authoritative database; here is the outcome

Both clearly have diagnostic service applications.  For example, 
combining the services of technical information presentation 
and failure mode analysis to determine capabilities when operating 
in a degraded mode (see Failure Mode Event Analysis).   As to 
the inference rules, if/then chains are typically used as a means 
to formulate possible outcomes.  

The semantic network has to be seen for what it is:  a set of typed 
assertions for typed data/concepts/objects.  It is one piece of 
the expert system network.   It's reliability, stability, coherence 
and stability is considered both in the local semantic net and 
more problematically, in using multiple nets in some given process. 

The services design using the BizTalk/SOAP framework can be used 
to enable the last but only if adequate control/tests are provided. 
For the ultimate applications of such, see the books on real time 
control systems as envisioned by the CASE design tool makers.  If you 
want a first pass at a process/control DTD, I wrote one for Beyond 
The Book Metaphor.  It is crude but it demonstrates the use of 
nested processes and control specifications for building stable 
cooperating systems.

Note:

"The problem with semantic networks for knowledge representation is still
that of 
ambiguity: there is an unlimited number of link and node types that may seem

appropriate, and their interrelationships will in general be very unclear.
In 
order to limit the set of types, we need an unambiguous, 
fundamental interpretation of what concepts and links in our network really 
stand for." 

"The meaning of a node is partially formal, determined by the network of 
semantic relations to which it belongs; and partially informal, determined 
by the personal interpretation of the user who reads the exposition, and
tries 
to understand the concept by associating it with the context. Such a format
allows 
the adequate representation of precise, mathematical concepts, of vague,
ambiguous, 
"literary" ideas, and of the whole continuum in between."


Len Bullard
Intergraph Public Safety
clbullar@ingr.com
http://www.mp3.com/LenBullard

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


-----Original Message-----
From: Dave Winer [mailto:dave@userland.com]

Wow, this has already been worth it. I suspected that a Semantic network
could be modeled as a hierarchy. I have been doing hierarchy editors (also
known as outliners) for a long long time. This is one of the reasons I'm
interested in getting a model up and running. I can think of it as a
thesaurus, one of my favorite writing tools, but what are the inference
rules about? How would we use inference rules on a semantic web of members
of the XML-DEV list? Dave