Adrian,
Thanks for the initial response on the
broad issue of comparing EE and CLCE. However, I believe you are mistaken about
two out of three of the differences you cite. Recursion and aggregation are
easily defined in First Order Logic (FOL) axioms. So, they should be well supported
by CL and CLCE.
I think you’re right about closed
world negation, though. I consider this a shortcoming of Common Logic (CL). More
generally I’ve heard that CL has no explicit support for nonmonotonic (or
default) reasoning, which I take to limit its utility in modeling commonsense information
and reasoning. But, CL does have some “context” mechanisms
that may help with that. And, it has true (open world) negation (as you
observe), which seems useful in any semantically rich environment. Although
closed world negation by failure works fine in many database environments, it
is more problematic in more expressive contexts (e.g., FOL) where rules or
axioms admit of a never-ending search for a failure to prove a conclusion. Even
if it doesn’t take forever to prove some negation by failure, the compute
cycles wasted can be considerable, which seems a practical reason for including
true negation in semantically rich information systems.
There is also, of course, the oft-cited reason
for supporting true negation in the Semantic Web: when dealing in an “open”
system like the web – you cannot assume that something is false simply
because you can’t prove that its true, and you do want to distinguish
between merely being unprovable and actually being false.
While I look forward to learning more
about the strengths and weaknesses of EE (especially for constructs like
propositional attitudes, tense, temporal constraints, causal relations, and quantifying
into modal contexts), it is probably best to take this off-list as you suggest.
It seems that this group wants to stay more focused on higher-level issues of
conceptual reconciliation/translation/communication between multiple systems
based on different domain ontologies than on the specifics of which ontology languages
(and language features) are best for promoting such interoperability.
Brian
__________________________________________
Brian A. Haugh, Ph.D.
Institute for Defense Analyses
Information Technology and Systems Division Phone: (703) 845-6678
4850 Mark Center Drive
Fax: (703) 845-6848
Alexandria, VA 22311-1882
Email: bhaugh@xxxxxxx
From: Adrian Walker
[mailto:adriandwalker@xxxxxxxxx]
Sent: Thursday, November 23, 2006
11:45 AM
To: Haugh, Brian; common upper ontology working group; Schoening,
James R C-E LCMC CIO/G6
Subject: Re: [cuo-wg] An Evolved
Approach to CDSI / Executable English
Hi Brian --
Many thanks for your note.
You wrote...
...using natural language to communicate
with the users/SMEs certainly seems essential. Previously, this has been
accomplished via conversation between the techie modelers and the SMEs. But, restricted [emphasis
added] English interfaces, like yours and others, offer the
promise of more direct and less ambiguous communication between SMEs and the
formal information model/ontology tools.
Agreed, English interfaces offer a lot of promise. However, the
Executable English interface supported by the Internet Business Logic system
[1] is actually an unrestricted
one. The technology is different from that of CLCE [2] and similar
systems. The vocabulary in [1] is open, and so -- to a very large extent -- is
the syntax. There is no need for external dictionary or grammar
construction or maintenance, yet the English semantics are strict. Since
English is at best a moving target, this approach has many practical
advantages. For example, it is not necessary to train authors or users to
use a restricted subset of English.
In [1], the unrestricted vocabulary and largely open syntax, together with
strict English semantics, is supported via a trade off. The idea is to
keep the natural language component as simple as possible, while
supporting SMEs and end users as they write executable content into
a browser, using their own words and phrases. I'll be glad to discuss
off-list how this
works. Hopefully, folks who would like to evaluate whether the trade off
is useful for their purposes will go hands on on the system to see for
themselves that it
works.
You wrote also...
I've been wondering if you have done a
comparison of your Executable English compared to the Common Logic Controlled
English (CLCE) form, especially with respect to expressive power, where CLCE
seems quite strong.
Yes. Here's an informal comparison.
Since CLCE [2] maps to FOL, I believe it lacks:
A. Recursion (needed for reasoning over
hierarchies -- e.g. transitive over [3]),
B. Aggregation (find the rolled up
available funding from all departments [4])
(find the total oil supply available [5])
C. Closed world negation (if Adrian is not in the list of employees of
IDA, then Adrian does not work for IDA).
[1] supports these. It also supports English explanations of reasoning
that involves A, B, and C.
CLCE [2] supports open world negation, which is currently considered to be
advantageous for the Semantic Web. However, in practice, most databases
are used via closed world negation.
[1] does not support open world negation, but could be extended to do so, if a
really strong practical motivating example could be found.
I hope this is helpful, and not too much technical detail. Do you
think there is any way of boiling this kind of discussion down so that it
would be of interest to Jim for the paper?
Thanks, -- Adrian
[1] Internet Business Logic, online at www.reengineeringllc.com.
Shared use is free.
[2] Common Logic Controlled English (CLCE). http://www.jfsowa.com/clce/specs.htm
[3] www.reengineeringllc.com/demo_agents/TransitiveOver1.agent
[4] www.reengineeringllc.com/demo_agents/FundManagement1.agent
[5] www.reengineeringllc.com/demo_agents/Oil-IndustrySupplyChain1Sqlmysql.agent