new issues - informal list

My understanding of the telecon discussion was that there was a desire to
have a much more comprehensive issue list than we currently have, with the
goal that the vast majority of our substantive e-mail traffic would be about
open issues.

Hence, as a brain storming exercise, I am going back other my contributions
to the e-mail list, and identifying the issues that concerned me. (and then
some more).

I am not formally raising any of these right now; but once that process is
ready, I will do I,J,Q,U,V. I would rather not raise all the rest, mainly
'cos it's a bit boring, and most are not really *my* issues, just issues
that I have some interest in. Also some are probably duplicates. If after a
week there are some that still look to me as if they need raising, then I
will do so.


Jim/Guus/Mike is this the sort of thing you wanted. I appreciate that
somewhat more detailed summaries of the issues are needed.



A: class theoretic entailments

which class theoretic entailments should hold in OWL?

B: dark triples

does OWL need an unasserted triple mechanism?

C: daml:collection != rdf:Bag

Why does DAML not use an RDF container.
Do the fixes of RDF Core allow OWL to use them.

D: daml:collection not a qname

The daml specs say that in rdf:parseType="daml:container" that the bit
before the colon is spelt d a m l. It however looks like a qname.

E: named or nameless class expressions

Is this an appropriate conceptual division. Should it be one characteristic
of the level 1, level 2 distinction.

F: xslt program from presentation syntax to exchange syntax

G: encoding ordering of properties and classes within a "frame" in triples

H: antifoundation

rdfs:Class rdf:type rdfs:Class

RDF has antifoundation, DL is well-founded.
Is this a problem? If so, what is the resolution.

I: datatypes - global idiom
(the XML value string is interpreted using a global range constraint)
DAML assumes untidy literals, RDF has tidy literals.

J: datatypes - local idiom
(the XML value string is interpreted using a global range constraint)
DAML uses types, RDF uses properties.

K: axiomatic vs model theory - normative status

L: axiomatic vs model theory - differences

M: taking subclasses and subproperties of keywords

It is not clear that this is legal according to the model theoretics
semantics of DAML+OIL.

N: can classes be members of classes

O: Do literal values and resources need to be disjoint

P: input completion and defaults

Q: There is no URI for xsd:string

R: unambiguous datatype properties

S: OWL ontology for OWL

T: divergence between idiomatic examples and formal semantics

e.g. a restriction can have two onProperty arcs but it would be somewhat
surprising based on the walk-thru and reference

U: charmod conformance of exchange lang

V: charmod conformance of presentation langs


W: ability to identify the properties and classes in an ontology

How are you meant to do this in DAML?

X: ability to identify a sub ontology

Y: versioning in DAML looks inadequate

Z: easy way of saying a load of names are distinct

a: uncle




Jeremy

Received on Friday, 19 April 2002 10:00:58 UTC