Re: ISSUE-52 (Explanations): Specification of OWL equivalences and rewriting rules for explaining inferences

From: "Kashyap, Vipul" <VKASHYAP1@PARTNERS.ORG>
Subject: RE: ISSUE-52 (Explanations): Specification of OWL equivalences and rewriting rules for explaining inferences
Date: Sat, 3 Nov 2007 13:28:41 -0400

> Clarification on this ISSUE-52"
> 
> [I] [w]ould like to propose that constructs for the following "extra-logical ?"
> features be included in the OWL 1.1 Spec
> 
> 1.	Proofs - probably as a sequence of entailments
> 2.	Entailments
> 3.	Explanations - probably as a human readable verbalizations of
>       entailments
> 
> The use case for the above is the need for an explanation feature for developing
> and debugging large scale OWL ontologies.
> 
> A standardized specification of the above would enable better tool support for
> these features enabling sharing of explanations
> across tools and applications.. This will increase productivity of the ontology
> developer.
> 
> [I] [w]ould like to discuss how this ISSUE is viewed as being within or beyond the
> scope of the current WG.
> 
> ---Vipul

What tools would generate these (potentially very large) constructs?
What tools would consume them?  

How are you going to account for variances between proof strategies
within a particular reasoning technology or variances between different
reasoning technologies?

What are the characteristics of these constructs?  For example, I can
think of a whole spectrum of specifications of proofs, ranging from ones
where a proof could always be "The ontology just entails the answer" to
ones that have the effect of nailing down exactly how an OWL reasoner
must behave.

What evidence do you have that the "explanations" can be divorced from a
particular UI context, and thus suitable for transmittal between tools
at all?


In general, WGs have to have one or more starting points - reasonably
well-worked out proposals that have a community already.  What is the
starting point here?  If there is none, then how can we proceed?

peter

Received on Monday, 5 November 2007 13:55:43 UTC