ontolog-forum
[Top] [All Lists]

Re: [ontolog-forum] Semantics1, 2, and 3 (Was: Summary on language and

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Internet Business Logic <ibl@xxxxxxxx>
Date: Thu, 13 Apr 2006 08:09:40 -0400
Message-id: <443E3F84.9020003@xxxxxxxx>
John --

Thanks for taking the time to comment (below) on the lightweight approach to natural language in our system.

I see from your web site that your work includes research to construct a general linguistically motivated ontology for interfacing between all computational components developed within our research centres and natural language technology components for natural dialog.

As mentioned, what our system does is counter-intuitive to folks working on dictionaries and grammars.  For example, the system would happily reason with your phrase

             English-as-the-natural-language-spoken-here-and-there-by-lots-of-people

How would you go about including that in your natural language technology components for natural dialog ?

What the system does is certainly much less ambitious in the NL area than your research.  Putting aside questions of whether what it does is "true English", or "just syntactic sugar", perhaps we could focus on whether it does something quite different from your research, something that is also useful? 

As mentioned previously, I have worked in dictionary-grammar NL research.  It's great fun.  This system is in no way a contribution to that area.   But it does appear to fill a niche in practical, robust, dictionary and grammar maintenance-free task specification and question answering.  One thing one can use it for is to reason over ontologies [1].

Please take these comments constructively -- they are so intended.

                                                                       Thanks,   -- Adrian Walker


[1]  http://www.reengineeringllc.com/demo_agents/MergeOntologies1.agent
      http://www.reengineeringllc.com/demo_agents/OntologyInterop2.agent
      http://www.reengineeringllc.com/demo_agents/SemanticResolution1.agent
      http://www.reengineeringllc.com/demo_agents/RelBioOntDefn3.agent
      http://www.reengineeringllc.com/demo_agents/FeaReferenceModelOntology2.agent
-- 

Internet Business Logic (R)
Executable open vocabulary English
Online at www.reengineeringllc.com
Shared use is free

Reengineering,  PO Box 1412,  Bristol,  CT 06011-1412,  USA

Phone 860 583 9677     Mobile 860 830 2085     Fax 860 314 1029


John A. Bateman wrote:
I didn't want to get drawn into this, however...

Adrian Walker wrote:
Actually, although the support for English in the system is technically rather simple, the approach means that the executable English that one can usefully write is essentially un structured. The
 vocabulary is open, and so to a large extent is the syntax. This, of
 course, is puzzling if one is used to classical dictionary-grammar approaches.  There is a trade off involved -- if an author wants the system to regard two sentences as having the same meaning, he must write that down explicitly.  An upside is that the English meanings are precise.

There is no relationship between the syntactic sugar dressed up
variety of a formal language that you are using and *English*,
i.e.,
English-as-the-natural-language-spoken-here-and-there-by-lots-of-people.
The only connection is in the eye of the beholder: i.e., the
human interpreter chooses to make a connection between the
kinds of restricted forms that are allowed to them and the
forms that they tend to use in their own language use. Since
you do not have an account of grammar that is compatible
to what is known of English, nor of the semantics that is
known for natural languages including English, nor for the
pragmatics and discourse structures that are used in natural
languages, any other claim verges on mispackaging. I think
it is important to be explicit about these issues just so
that it is clear when one is engaging in metaphor (valid
for sales purposes perhaps) and when one is doing theory,
either linguistics or ontology.

Actually, although the support for English in the system is technically rather simple, the approach means that the executable English that one can usefully write is essentially un structured.

If what can write is unstructured, then this is also a major
deviation from natural language, in that what one can say
or write in natural language *is* structured, and highly
so. To understand this, look at any text written by a
half-proficient non-native speaker. The structure violations
at all levels of description are usually striking.

Even the term:
for executable English content

is in danger of going round the loop of confusing
ontology, formalization and language again, just when things
were getting clearer. Replace "English content" with "pretty
printed formal representation with syntactic sugar", then
we'd all know both what was being done and that you know
what was being done! Of course, it is more difficult to sell
that way: but it is just *not* something that bares a natural or
simple relationship to natural language.

The vocabulary is open, and so to a large extent is the syntax.
This, of course, is puzzling if one is used to classical
dictionary-grammar approaches.

:-) It is also puzzling when one is used to working
with natural language!

John B.



_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Subscribe/Unsubscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx





<Prev in Thread] Current Thread [Next in Thread>