On Thu, Feb 14, 2008 at 10:23 PM, Mike Tintner <[EMAIL PROTECTED]> wrote:
> Vlad: Article is about problems with deductive-style ontologies, so it
>
> > should apply to Cyc, but not to NARS. Dealing with different levels of
> > belief (evidence) and context-sensitivity of concepts is central to
> > NARS
>
> Vlad,
>
> Thanks for reply. The central criticism of the article for me is that
> logic - syllogisms - on the whole produce trivial results. Why are the
> deductions that all Greeks are mortal, or that Brooklyners speak with
> Brooklyn accents, any less trivial because they are accompanied by levels of
> belief, ("well I'm not terribly sure about that"), or context-sensitivity,
> ("of course it all depends what you mean by "mortal" or "accents")? Seems to
> me - triviality is triviality however you dress it up. And I have to say I
> have never (in my admittedly limited experience) seen other than trivial
> results from AI logical reasoning. Have you? Could you give an example or
> two?
>
> And what has logic got to do with AGI - i.e. problemsolving in unfamiliar
> domains, where, by definition surely, logic, (which can only work on
> familiar, formal knowledge), cannot apply?
>
'Logical AI' is a blurry concept. It's possible to call many
sufficiently small algorithms that work with data records 'logics'.
Probability theory is a logic. NARS is a logic. So this concept is a
wrong tool in this case.
As I see it, the main problem can be characterized by excessive
restrictions on modes of reasoning that can be applied to data that is
presented to inference engine. If ontology comes with specific
semantics of its relations, it doesn't actually represent concepts
that it talks about, and subsequently problem that is being solved by
devising various inference procedures is not one that we originally
want to solve, that is to make a system that is able to understand
real world.
--
Vladimir Nesov
[EMAIL PROTECTED]
-------------------------------------------
agi
Archives: http://www.listbox.com/member/archive/303/=now
RSS Feed: http://www.listbox.com/member/archive/rss/303/
Modify Your Subscription:
http://www.listbox.com/member/?member_id=8660244&id_secret=95818715-a78a9b
Powered by Listbox: http://www.listbox.com
