Re: Joint Committee Minutes 3 April 2001

From: pat hayes (
Date: 04/09/01

Some tag questions.

But first, a grousing about terminology:
Jim Hendler.....
Higher Order Logics

You know what these are - key idea is that logic statements can refer to
other logic statements, i.e. (Believes John (believes Fred (believes Sam
(Isa WebHacker Dan)))).
No and no!  First, higher-order isn't the same as meta. HO logic 
quantifies over *relations* (and relations on relations, and ....). 
Those relations are things in  the semantic domain, not statements. 
Theres nothing in HO logic that has statements referring to other 
STATEMENTS. That would be a meta-language (which KIF 3.0 tries to 
provide, for example, though we now think they had it a little 
screwed up), which is a different thingie altogether.
And in any case, the example given here involving Believes is yet a 
third idea, ie a modal logic. Modal logic is not higher-order at all 
(though I guess you could have a higher-order modal logic, since they 
are kind of orthogona) and they aren't metalanguages either. 
(Believes John (Isa WebHacker Dan)) is NOT the same as
(Believes John '(Isa WebHacker Dan)), and it can't be properly 
interpreted that way. (That is what Montague showed a while back, 
that the 'substitutional' interpretation of the modalities doesnt 
work.)  Modal logics can be translated back into regular non-modal 
logic (vanilla first order) which is why I personally think that 
modalities are a red herring, but others disagree with me here. But 
certainly they arent the same as higher-order or as meta-languages.

OK, end of grouse, back to tagging.

Jim also referred to 'annotated logic', about which I knew nothing, 
but I found an account at (you have to scroll 
down to find it, its in:
S.M. Leach, Bucknell University, Lewisburg, USA. On Computing
Annotated Logic Programs. )
As far as I can see, what this is about is using 'annotations' which 
are bounds of the truth-values of atoms in some truth-value lattice, 
ie this is a technique for reasoning with partial or probabilistic 
logics. I don't quite see how that helps us with the kind of tagging 
that we need... or am I missing something here, or have got the wrong 
notion of 'tagging'?


I have a few basic questions about tagging. Obviously it is a good 
thing to let folk write post-it notes on things, for who knows what 
reasons they might have. But I get the sense that this notion of 
tagging goes beyond just a kind of sophisticated commenting or 
documentation facility, and is supposed to get into the ontology 
content itself in some way.

First question: can a tag alter or modify the logical force of any 
sentence? In other words, if I assert P plain, or if I assert P with 
a tag T, does the presence of T alter the assertion of P itself, or 
is it simply something else added to (conjoined with?) P?

Second question: assuming the above answer is 'no', so that adding a 
tag only adds to the information in the assertion (ie in a sense, 
tags are monotonic (which annotations in ALP are not, apparently, by 
the way)), presumably we can think of the asserting of P tagged with 
T as meaning the same as a conjunction of P with something else. If 
we do that, what is the something else? Is it T itself, or some kind 
of compound made up from P and T together?

Third question: does a tag *refer to* the sentence it is tagging? Ie, 
are tags in a metalanguage?

Pat Hayes

IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola,  FL 32501			(850)202 4440   fax

This archive was generated by hypermail 2.1.4 : 04/02/02 EST