Pat, (01)
I mostly agree with your comments, but I'd like to clarify
some points. (02)
JFS>> During the development of the Common Logic standard,
>> Lenat had been asking for the kinds of extensions to CL
>> that were added to IKRIS. (03)
PH> Actually the pressure came mostly from the Stanford IKL
> laboratory, who were completely sold on the idea of context
> logics, influenced by John McCarthy. The Cyc participants
> in IKRIS were quite happy to translate their contexts into
> classical FOL if necessary. (04)
I agree with that point about who initiated the IKRIS project
and what the Cyc representatives to IKRIS said. But I was
talking about points that Lenat had been saying since the
1990s about the kind of logic he needed as a target for
translating full CycL. (05)
As you know, the requirement for something along the lines
of McCarthy's contexts to support CycL is not surprising,
since John McC. was Guha's thesis adviser and one of the
chief influences on getting those features into CycL. (06)
PH> IKL (the IKRIS extension) doesn't add contexts to the logic,
> but rather adds proposition names. These can then be used
> to emulate context reasoning in a non-contextual logic. (07)
I agree. The old KIF 3.0 had a backquote, which I used to
support contexts in conceptual graphs (which were influenced
by Peirce's contexts, which have similar expressive power
to McCarthy's contexts). I was very happy to get the 'that'
operator in IKL, which can be used in the same way as backquote. (08)
PH> But its reasoners can run very fast and are guaranteed to
> be decidable, which is the reason for accepting the limitations.
> That at any rate is the official line. (This all refers to OWL-DL,
> not OWL-Full.) (09)
I agree that is the official line. But people who translate OWL
to Prolog support very large knowledge bases with better speed
and industrial-strength reliability than current OWL software.
(In my previous note, I mentioned Experian, which has huge
databases and enough money to buy the entire Prologia company.) (010)
Furthermore, decidability for anything the size of the WWW is
a crock of "euphemism". Even polynomial algorithms don't
scale, and decidability on the kind of data found on the WWW
is decidedly unrealistic. (011)
JFS>> My recommendation is to adopt a single, very general
>> semantic representation and to relate the multiple
>> specialized languages to the common semantics. (012)
PH> I agree. But this is not a popular strategy. (013)
I certainly agree. That is why anybody who maps the SemWeb
notations to a more general logic has a superb advantage.
At VivoMind, we are happy to recommend OWL to our competitors. (014)
John (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)
|