Response to JanM1

From RIF
Revision as of 23:28, 28 September 2008 by Harold Boley (Talk | contribs)

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

Jan Maluszynski wrote:

> In my opinion the Working Draft on RIF-BLD of 30 July 2008
> is a  mature  document and I would be glad to see  a Candidate
>  Recommendation based on it. My interest on RIF-BLD is related
> to my work on integration of rules and ontologies under the
> well-founded semantics. 
> (see e.g. Wlodzimierz Drabent, Jan Maluszynski: 
> Well-Founded Semantics for Hybrid Rules. RR 2007:LNCS 4524,  1-15).
> 
> We are also working on rule languages
> where uncertainty is handled in the framework of Rough Sets.
> (see e.g. Jan Maluszynski, Andrzej Szalas, Aida Vitria:
>  A Four-Valued Logic for Rough Set-Like Approximate Reasoning. 
> T. Rough Sets 6: 176-190 (2007))
> 
> Some comments:
> 
> - Having more examples would be very helpful,

We are working on a separate document, which contains many test cases (http://www.w3.org/2005/rules/wiki/Test). This document has numerous examples along with explanations of what should be inferrable from which RIF-BLD documents (http://www.w3.org/2005/rules/wiki/Category:Test_Case).

> - The direct specification of RIF-BLD Semantics  (Section 3)  
> follows closely the Semantic Framework of the Working Draft on RIF-FLD.
> This is good,  but I would also expect to see  a  specialization of the
> general framework to the specific case of RIF-BLD.

We don't quite understand this comment. Section 3 is rather prominent in the RIF-BLD document, and it deals with this issue.

> In particular, as RIF-BLD is claimed to correspond to the language
> of definite Horn rules, the minimal Herbrand model semantics, should
> perhaps be discussed.

There is no need to discuss multiple semantics in the *Basic* Logic Dialect, especially since the Herbrand semantics is less general. Logic programming is not the only planned extension of RIF-BLD that the group is considering. A first-order logic dialect is also being contemplated.

> The minimal Herbrand model is   mentioned in
> Section 3.8 of the RIF-FLD Working Draft as  an intended semantic
> multi-structure of a RIF-BLD Sets of formulas,
> but is not mentioned at all in the RIF-BLD working draft. 
> Having clearly defined Herbrand models for RIF-BLD is
>  important for the extensions where rule bodies include
> negation-as-failure and for the approaches aiming on hybrid  integration
> of such rules and ontologies (including our work based on the well-founded
> semantics).

So, the dialects in question should define this special type of RIF semantic structures.

> - The presentation syntax seems to be very useful. Having in addition 
>  a variant of it using standard mathematical notation for quantifiers
>  would be convenient in some cases, like teaching students with 
> the mathematical background.

The presentation syntax is designed to simplify writing the semantics and examples. RIF is not a teaching tool, but a specification of a future standard. In teaching, one can use whatever syntax is suited best.

> Also, I wonder if the  universal 
> quantification of rules must be explicit, in contrast to the standard
> implicit quantification used in logic programming.

There was a disagreement on this issue in the WG. Being explicit is a principle of markup syntaxes. Ultimately, this design decision does not matter for the most part, since RIF is an exchange format, not an actual language for authoring rules. Furthermore, many WG members see the main value of RIF in the dialects that extend RIF-BLD (such as logic programming dialects) or extensions of a core subset of RIF (such as the production rules dialect). These dialects are not bound by the design decisions imposed by RIF-BLD.

> Jan Maluszynski
> Department of Computer and Information Science 
> Linkoping University
> 581 83 Linkoping
> Sweden 


Thank you very much for the useful comments.


-The RIF WG