[Top] [All Lists]

Re: [ontolog-forum] Foundation ontology, CYC, and Mapping

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Len Yabloko <lenyabloko@xxxxxxxxx>
Date: Tue, 16 Feb 2010 08:36:46 -0800 (PST)
Message-id: <522701.34111.qm@xxxxxxxxxxxxxxxxxxxxxxxxxxx>
Dear John and PatC,     (01)

I would like work out some fundamental principles that we all can agree on: 
> Some fundamental principles:
>   1. A computer cannot do anything with "intended
> meanings" in the
>      head of some programmer or some
> human being who runs the program.
>     (02)

This is true of any intelligent agents (human or not). And yet humans are much 
better at cooperating based on intended meaning. I don't believe our brains 
operate on intended meaning though. The operative scope is broader than that - 
we operate on the basis of context and try to infer the intended meaning, and 
then validate our inference. Why can't computers do the same in principle?          (03)

>   2. The only meanings that are relevant to the
> computer are the ones
>      that are embodied in the programs
> that the computer runs.    (04)

That would correspond to strictly deductive reasoning. But we all know that 
deduction is not sufficient for bulding an intelligent agent. If we include all 
cooperating agents in the loop, then logic can be applied to information 
outside the programs such as inputs and outputs.    (05)

>   3. Those meanings must be derived from some kind of
> specifications,
>      which are translated into
> executable machine code either by a
>      human programmer or by some
> compiler that automatically translates
>      a formal specification into
> executable code.
>     (06)

This principle only applies to the intitial context of agent's interaction. As 
soon as new information is obtained in the course of the interaction the 
meanings can be derived inductively and mutualy tested. I think this is what 
happend when human interact.    (07)

>   4. If we want to use an ontology to ensure
> interoperability those
>      specifications must be so
> precisely defined that any two coders
>      (human or machine) will generate
> equivalent machine code.    (08)

I diagree. First of all how precisely is "so precisely" can not be determined 
up-front. In software this is known as "late binding". And second - there is no 
need to generate equivalend code - only to generate compatible code. Encoding 
the information for remote interpretation is not necessarily the main objective 
of interaction, or even a prerequisite in some cases.      (09)

>   5. Words like 'primitive' are so vaguely defined
> that they provide
>      little or no guidance to
> programmers.  Pat H. and I have been
>      trying to explain the formal
> relationships between specifications
>      and machine code.  Arguments
> over the meanings of words like
>      'primitive' are irrelevant.
>     (010)

I would say that formal relationships between the specs and the code do not 
cover the issue at hand. This is why words like 'primitive' keep coming up. In 
my view 'primitive' simply stands for unit of context. I relaize that is not a 
definition. But executable code does not server that purpose too. What we must 
specify is a set of constraints which can not be reduced to executable code, 
but rother allows possibly infinit number of ways to satisfy. I would like 
someone on this forum to formaly state that (if it is possible at all).     (011)

> PC> Given these two interpretations of "primitive" in a
> **mathematical**
>  > theory, it seems that the "meanings" of terms
> (including primitive
>  > terms) in a mathematical theory have little
> resemblance to the
>  > meanings of terms in a computational ontology that is
> intended to
>  > serve some useful purpose...
>     (012)

I agree with Pat C. And I hope it is clear why from the above comments.    (013)

-Len    (014)

Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com     (015)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: 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 join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (016)

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