Re: OWL-S: simple proposal to evolve logic expressions

David Martin wrote:

>
> Here is a simple proposal for evolution of logic expressions used in 
> OWL-S (see
>   http://www.daml.org/services/owl-s/1.1/generic/Expression.owl.)
> These changes would appear in release 1.2.
>
> This is to allow for unquoted SWRL and DRS expressions, and also to 
> add some new categories of expression (SWRL-FOL, RDQL and SPARQL), as 
> discussed in recent telecons and private messages.
>
> OWL-S folks (and interested parties), please review and comment if 
> needed.
>
> --- Background
>
> Currently we have class Expression with subclasses
>     SWRL-Expression, DRS-Expression, KIF-Expression
>
> Expression has DatatypeProperty expressionBody (the values of which 
> are the actual "quoted" expressions), with range restrictions:
>   for SWRL-Expression range = XMLLiteral
>   for DRS-Expression range = XMLLiteral
>   (not restricted for KIF-Expression, but understood to be a string)
>
> --- Proposed change (1):
>
> Add new subclasses of Expression:
>     SPARQL-Expression, RDQL-Expression, and SWRL-FOL-Expression
>
> --- Proposed change (2):
>
> Add a new property of Expression:
>   ObjectProperty expressionContent (for "unquoted" expressions)
>
> expressionContent will be used with SWRL-Expression and DRS-Expression
>
> expressionBody will be used with KIF-Expression, SPARQL-Expression, 
> RDQL-Expression, and SWRL-FOL-Expression
>
> --- Proposed change (3):
>
> Mandate the above conventions with cardinality restrictions:
>
> Specify cardinality = 0 on expressionBody and cardinality = 1 on 
> expressionContent on SWRL-Expression and DRS-Expression
>
> Specify cardinality = 1 on expressionBody and cardinality = 0 on 
> expressionContent on KIF-Expression, SPARQL-Expression, 
> RDQL-Expression, and SWRL-FOL-Expression
>
Clever!

> --- Proposed change (4):
>
> Restrict range of expressionContent:
>   for SWRL-Expression range = union of Imp, AtomList, and Atom

Not Imp, I think. Imps don't "evaluate" to true or false (they're not 
quite implications in the "logical connectives" sense). In fact, it may 
be easier to just do AtomLists (if there's a single Atom, then we just 
have a list with one element).

>   for DRS-Expression range = <whatever Drew says>
>
> Thanks,
> David
>
>

Received on Thursday, 19 May 2005 16:39:05 UTC