Response to CD2

From RIF
Revision as of 03:37, 28 November 2009 by MichaelKifer (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Thank you for the suggestion. It has been adopted and the semantics of :- has been modified per your suggestion.

> Dear all,
> The semantics of the several connectives in FLD-RIF is many-valued,  
> with the single exception of the rule implication symbol.
> The rule implication symbol has only two possible values, t and f, the  
> top and bottom of the lattice of truth-values.
> For instance, this prevents the definition of fuzzy logic dialects of  
> RIF without introducing new implication connectives.
> I suggest that that the rule implication symbol semantics to be  
> generalized in the following way:
> Rule implication:
> TValI(head :- body) = t, IFF TValI(head) ≥t TValI(body).
> TValI(head :- body) < t   otherwise. 
> This has nice theoretical properties, namely the existence of minimal  
> model for every definite Horn program.
> This will not affect the notion of model, and would be much more  
> general.

Thank you for the suggestion. It has been adopted and the semantics of :- has been generalized per your suggestion.

> Disclaimer:
> I've tried to browse the mail archives to see if there was any  
> discussion about this, and could not find it.

No, there was no prior attempt to generalize :- in the direction that you suggested.

> I don't discuss here the issue of equality, since I am not an expert  
> on the subject.

The semantics of = is two valued. It is possible to generalize it and make equality multi-valued, but perhaps this is better done as a separate predicate (which FLD allows dialects to have). Your and others comments regarding this issue are welcome.