W3C home > Mailing lists > Public > www-webont-wg@w3.org > February 2003

Re: syntax task force - differences between the two approaches

From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
Date: Fri, 28 Feb 2003 14:45:05 +0000
Message-ID: <3E5F75F1.1000205@hpl.hp.com>
To: "Peter F. Patel-Schneider" <pfps@research.bell-labs.com>
CC: www-webont-wg@w3.org



Peter F. Patel-Schneider wrote:

> Here is my summary of the differences between the two approaches.  I may be
> missing some differences.
> 
> peter


Here's my attempt at grouping


Resolved:
> Jeremy - forbids unused owl: vocabulary (but not unused rdf: rdfs: or xsd:
> 	 vocabulary) 
> S&AS   - allows any unused vocabulary
> 


Stylistic differences that are not important:
> 
> Jeremy - all names in abstract syntax need tags
> S&AS   - names in abstract syntax are not tagged
> 
>
> Jeremy - syntax includes side condition on non-simple properties not
> 	 allowed in cardinality-restricting constructs
> S&AS   - condition is a side condition

(I find both my choices easier to work with, but Peter's choices make 

a document that is easier to read,  at least for some readers).

(With the side condition I think that Peter's newer text is a
significant improvement).





> 
> Substantive Differences in Abstract Syntax
> 

 >

Important ones:
 > Jeremy  - all names need rdf:type triples
 > S&AS    - ontologies and annotation properties do not need rdf:type triples
 >

 > Jeremy - only binary equivalence and disjointness for classes
 >          (not a semantic restriction, of course)
 > S&AS   - n-ary equivalence and disjointness for classes
 >
 >
 > Jeremy - forbid complex single-property restrictions
 > S&AS   - allow complex single-property restrictions
 >
Lesser important ones:
 >
 > Jeremy - annotation properties (but not ontology properties) have a
 > 	 declaration that can have annotations
 > S&AS   - no declaration for annotation properties
 >
(I can drop annotations of annotations but allowing a declaration of an 
annotation property is "lesser important")
 >
 > Jeremy - top-level unnamed descriptions (and restrictions) allowed in
 > 	 abstract syntax
 > S&AS   - unnamed descriptions (and restrictions) can only occur inside
 > 	 other constructs in the abstract syntax
 >

Technical details:

> Jeremy - all ontology information in an abstract ontology is in a header
> 	 construct 
>        - allows imports, etc, for multiple ontology resources in a single
>          ontology 
> S&AS   - abstract ontologies have a (single) optional name
>        - all imports, etc. work off this name (or an unnamed resource)
> 

(when I looked at it S&AS imports did not work, which motivated change; I 
haven't yet analysed peter's latest offerings here)


>
> Jeremy - non-DL properties (properties that are neither object or data
>          properties) are divided into annotation properties and
>          meta properties (should instead be ontology properties)
>        - annotation properties can only relate to individuals and data
>          values
>        - ontology properties are a fixed, predefined set
> S&SA   - non-DL properties are not sub-divided
>        - non-DL properties can relate to any resource
> 

Motivation:
S&AS appears to allow

Class(<a> annotation( <owl:backwardCompatibleWith> <b> ) )

since owl:backwardCompatibleWith is intended to relate ontologies it should 
have domain and range of owl:Ontology and hence cause confusion in the 
above example.
backwardCompatibleWith does not appear in owl.owl ...
owl:imports is not given a domain and range in owl.owl ...
(there is a note in the current ref WD to the effect that these will be added).



Substantive Issuettes:

(i.e. I think my solution is better, but this is orthogonal to 

the general point of syntactic tidiness; and these can be dealt 

with independendently of the other differences, and of each other)
> Jeremy - can name data valued oneOfs
> S&AS   - can't name data valued oneOfs
> 
> Jeremy - incorporates some RDF container vocabulary
> S&AS   - forbids RDF container vocabulary
> 
> Jeremy - allows rdf:XMLLiteral
> S&AS   - forbids rdf:XMLLiteral
> 

(slightly bigger than an issuette, but orthogonal)

> Jeremy - impossible to state some different/same patterns for unnamed
> 	 individuals 
> S&AS   - impossible to state any different/same patterns for unnamed
> 	 individuals 
> 
>



 
> 
> Bugs 
> 
> Jeremy - missing rdfs:seeAlso
>        - rdfs:comment has wrong category
>        - rdfs:isDefinedBy has multiple categories which is not supportable
>        - lots of grammar ambiguities (but only benign ones)
> 
> S&AS   - lots of grammar ambiguities (but only benign ones?)
>        - ....
> 
> 
Received on Friday, 28 February 2003 09:45:37 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:57:57 GMT