RE: NAF and NEG [was: LP Semantics (non-monotonicity) in Usage Scenarios?]

> Thanks for this listing of types of predicates.
> 
> I do have a question, however, on POSCLOSED.  You state that such
> predicates are "for use only in the current module," but then
> immediately need to use such a predicate outside the module.  You
> handle this by creating a redundant predicate which is VISIBLE
> along with rules for concluding the NEG of the predicate extent.

Right.
 
> Why not create VISIBLE-OPEN and VISIBLE-CLOSED to handle such cases?

Sure. This is a possibility. I was trying to clearly define the scope of
non-monotonic predicates and hide inside all the non-monotonic inferences.
But, surely one can define VISIBLE-CLOSED and VISIBLE-OPEN predicates; I
just did not want to propose that :-)

> A vendor may wish to have a number of predicates describing properties
> of its products.  It wants others to be able to use these predicates,
> but not to alter any asserted properties.  Instead of making dopleganers
> of each predicate with rules for populating them, it would seem far
> simpler to declare the predicates visible, yet closed.

There is also an important point regarding my closed predicates: you can use
rules to define closed predicates.  For instance,

listedMemberEU(?C) :- foundingMemberEU(?C).
listedMemberEU(portugal).
listedMemberEU(spain).
listedMemberEU(?C) :- memberEurope15(?C).
listedMemberEU(?C) :- enlargementEU2004(?C).

By adding the completion rule

NEG listedMemberEU(?C) :- SNAF listedMemberEU(?C).

You get the expected conclusions (if you define the auxiliary predicates
appropriately). If I understand correctly the "semantics" of log:notIncludes
cannot be used here to implement a similar behaviour. This is more in the
spirit of log:definitiveDocument.


> You raise such a possibility as possible.  There seems to me a strong
> need for such predicates.

Thanks. I'll take into account your comment in a subsequent version of my
draft.
 
Carlos

Received on Monday, 29 August 2005 15:31:59 UTC