Sean, Sergei, and Pat C., (01)
SB> For me a more interesting question is what can ontologies
> do reliably? and how reliably? (02)
I agree that is a more interesting question. I was trying to
discourage Pat C. from using the word 'understanding' and focus
on the issues of how we should design the computer system. (03)
SN> First, our main problem is well beyond choosing the "right"
> knowledge representation schema: it is all about the content
> of knowledge, not the format. (04)
I agree that the content is more important than the format,
but the kind of processing is also extremely important.
Although I think logic is important, I also believe that
induction, abduction, and analogy are at least as important
and probably more important than deduction. (05)
SN> Second, the scope of work is so much broader than even
> the most sober among us had expected. (06)
I agree. (07)
SN> But the bad news is that I think our problem is more
> complex than either of those problems [the Manhattan
> Project or the Human Genome Project]. (08)
Absolutely! Those are almost trivial by comparison.
In the 1960s, engineers thought that hardware design was
more difficult/important than software design. I would
compare the genome to the hardware. Although the amount
of information in the genome is large, it's tiny compared
to the amount of information in the brain. Furthermore,
the hardware/genome changes much more slowly than the
information in the software. (09)
SN> This is one of the reasons why I think that no standards
> could really be enforced in this area and that it may be
> a noble but doomed task to try to come up with a single
> common syntax and semantics for the metalanguage for
> specifying knowledge about language and the world (whether
> these are different metalanguages or a single one). (010)
That depends on how you define "this area". First-order
logic has been around for over 125 years, and Common Logic
is a minor extension of what Frege and Peirce independently
discovered. I don't think that's the solution to everything,
but my major complaint about the Semantic Web is that they
didn't build on the most successful application of FOL --
namely, relational databases with SQL (or better, Datalog). (011)
As far as ontology goes, I have been arguing *against* any
fixed upper level and in favor of standards that support ways
of accommodating multiplicities of ontologies. Alan Bundy,
who has been working with logic for years, strongly agrees. (012)
JS>> Or would it be better to put all the knowledge about bridge
>> in a module that deals with bridge and all the knowledge about
>> tomatoes in a module that deals with tomatoes? (013)
SN> Any which way. Let it even be inefficient. But we need multiply
> cross-indexed descriptions of complex events with their subevents
> and participants, pre- and post-conditions and other properties. (014)
I agree. But my main point was to avoid packaging the world
knowledge too tightly with the lexicon -- partly because it
would make it more difficult to reuse the knowledge with different
languages (both natural and artificial). (015)
SN> BTW, there are many more kinds of ambiguity to deal with in
> addition to word sense or PP attachment... (016)
Certainly. I just dragged out that example because it illustrated
a few points related to the word-expert issues. Word experts are
irrelevant to resolving those other ambiguities. (017)
SN> The organization in our approach is by (ontological) elements
> of world knowledge but our lexicon expresses lexical meaning
> in terms of the ontological metalanguage... So, in the example
> above, there will be in the ontology the event describing what
> happens when people play bridge, and there will be indications
> in the lexicon of any idiosyncratic word and phrase senses
> relating to bridge playing. Many meanings will still be derived
> in a compositional way, with the knowledge of the complex event
> of playing bridge serving as a (core) heuristic for making
> preferences during ambiguity resolution. (018)
As I interpret that paragraph, your ontological approach would
put world knowledge into the language-independent representation
and put language-dependent information in the lexicon. I would
endorse that approach. And, I believe, it is very different from
a word-expert parser that binds world knowledge with the words. (019)
It's important to use world knowledge during the parsing stage,
but I would recommend a dynamic method of combining the three
kinds of knowledge -- syntactic, lexical, and world knowledge
-- during the analysis stage. A static combination in "word
experts" would imply that domain knowledge encoded in an English
word expert would have to be rewritten for Russian or French. (020)
SN> However, the main issue is that we need to be able to make
> successful inferences against a knowledge base that is not
> sound and complete. That's reality. So, if logic can come up
> with methods that support such a task, great. Otherwise, we
> scruffies will have to make do with whatever we can muster. (021)
I strongly agree. But I view analogy as a "scruffy" way
of using logic. I consider formal deduction to be a special
case of analogy, as in the article by Arun and me: (022)
http://www.jfsowa.com/pubs/analog.htm
Analogical Reasoning (023)
PC> Those kinds of pragmatics are what would be included in the
> Word Expert, or in a 'topic expert' that would have a broader
> understanding of plants and gardening generally, not only tomatoes. (024)
I would be much happier with talk about topic experts than word
experts. A topic expert can be language independent and reusable
in applications that use different natural *or* artificial languages. (025)
PC> As best I can tell, Cyc's language interpreter does not use
> Word Experts. Big mistake. (026)
Topic experts would be great -- they could support Wittgenstein's
notion of language games, which I like very much. But word experts
would be a disaster, because they're the opposite of modularity.
There are three kinds of information to be represented: (027)
1. Language-dependent, but domain independent information
about grammar. (028)
2. Domain-dependent information about an open-ended
variety of subjects. (029)
3. Lexical information about words and their connections
to the grammar (point #1) and the domains (point #2). (030)
If you combine #2 and #3 in word experts, you get a combinatorial
explosion in the amount of *human* effort required. And you make
it impossible to reuse the knowledge in different languages. (031)
Modularity is extremely important, and that was the main point
of my talk at FOIS'06: (032)
http://www.jfsowa.com/pubs/dynonto.htm
A Dynamic Theory of Ontology (033)
John (034)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (035)
|