On 4/26/2011 6:40 AM, Mike Bennett wrote:
> There is a use case for decidable stuff, it's just much smaller
> than the use case for integration of legacy systems. (01)
Actually, I believe that the problem of decidability is very important
and every programmer should understand the basic principles. But even
programmers who have never studied the subject (or even heard about it)
have good intuitions that certain ways of programming can lead to
combinatorial explosions. (02)
But no programmers would ever dream of restricting the expressive
power of their languages. Instead, they have developed methods of
"structured programming" and "design patterns" that are known to be
reliable and efficient. (03)
That same principle works equally well with reasoning systems.
The CycL language, for example, is very expressive, but that
expressivity doesn't create any problems. What the Cyclists
have done is to define the equivalent of "design patterns" that
can be used both by the knowledge engineers who write CycL,
and by the reasoning tools that select which patterns to apply
for any particular problem. (04)
> I think it's a balance between solving old problems and carving
> out new opportunities (things which are not known to be a problem
> and so don't have a use case). Both are significant in the long
> run. We saw some good ones in the case studies. (05)
I certainly agree. That is my major complaint about the SemWeb:
they made no attempt to bring the mainstream into the new world
of semantics. I was hopeful about the SemWeb back in 1998, but
I have been very disappointed by their "provincial" views. (06)
> That's why I think it was so important that this Summit
> recognized the difference between different kinds of ontologies
> built for different use cases and using different architectures. (07)
Yes. OWL has some useful applications. But it is so constricted
that it prevents people from seeing the huge range of possibilities. (08)
In summary, restricting the expressive power of a language has
three kinds of effects: (09)
1. It makes it impossible to state many important problems,
some of which could be solved very efficiently in a more
expressive language. (010)
2. It forces the knowledge engineer to state many problems in
a grossly contorted way that are actually *less* efficient
than a more natural expression in a more expressive language. (011)
3. It blocks innovation by preventing kn. engineers from inventing
new design patterns for handling new kinds of problems. (012)
Mathematicians have understood these principles for centuries.
Many problems that are difficult or impossible to solve in
integers can be handled much more efficiently in real numbers.
Other problems require an extension to complex numbers. And
see the latest issue of _Scientific American_ (May 2011) for
a short article about octonions: "The strangest numbers in
string theory:," p. 60. (013)
Bottom line: Restricting expressive power can *never* solve
any problem more efficiently. It just makes some problems
impossible to state. (014)
John (015)
_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2011/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2011
Community Portal: http://ontolog.cim3.net/wiki/ (016)
|