RE: REQDOC: Need final input!

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Jonathan,
Trust/provenance issues are indeed important. I believe the editors
wanted to limit changes to small edits. The semantic web layer cake
suggests Trust is a capstone layer sandwiching rules/logic and
ontology. It may not be obvious what aspects of trust/provenance are
topical for the ontology layer.

Personally, I think DSIG went too far (although they had good
justification for doing so) in defining containing structures, but
this was done outside the context of RDF and OWL. 

OWL/RDF provide the framework for defining meaningful structures. The
DSIG containing structures may be superfluous in the context of OWL
defined structures. DSIG also dovetails into existing standards for
certificates (X.509, SPKI). Certificates are provenance data
structures that carry with them the baggage of pre-web infrastructure
(e.g. X.509 distinguished names). Web namespaces are a fundamentally
different approach to naming. Hence, identification and
authentication (I&A) mechanisms should be re-thought.

The really valuable contribution OWL could make in the context of
Web-based PKI is to provide the language for defining trust &
provenance data structures. It is important to recognize that such
structures dovetail into the ontology structure. Hence, should
leverage the native data description language (DDL) native. I assert
OWL/XML will be the DML/DDL for the Semantic Web.

X.509, as specified by X.500, used ASN.1 as the DDL for the X.500
data model. Certificates, not surprisingly used the same DDL and data
model as X.500. X.500 failed to become a global infrastructure
however. 

- From the Semantic Web viewpoint certificates and accompanying
provenance semantics should be expressed using the same DML/DDL and
data model adopted by the semantic web. That is Ontologies are the
data model. OWL/RDF? is the DML and XML is the DDL. The certificate
standard for the Semantic Web should leverage this infrastructure.
Hence some parts of DSIG may be superfluous given OWL to capture
container and provenance semantics.

Regards,
Ned


> -----Original Message-----
> From: Jonathan Borden [mailto:jonathan@openhealth.org]
> Sent: Tuesday, February 26, 2002 4:41 PM
> To: Smith, Ned; 'Jim Hendler'; Dan Connolly; Jeff Heflin
> Cc: WebOnt
> Subject: Re: REQDOC: Need final input!
> 
> 
> Ned,
> 
> >
> > Not to split hairs, but the wording in the last line of 
> Jeff Heflin's
> > text might raise some flags. "It should be straightforward" may
> > suggest we don't understand the basis of the interaction between
> > XML Signatures and OWL.
> >
> > I recommend wording to the effect of:
> > ...
> > "XML Signatures and cononicalization transformations can be
> > applied safely to OWL expressions because OWL conforms to the XML
> > standard." 
> >
> > - -Ned
> 
> What about the capability for OWL to deal with 
> trust/provenance issues? If
> anyone can say anything about anything (using RDF), we need a 
> way to decide
> what to reason about, that is, assertions might be filtered 
> by XML sigs etc.
> I certainly don't understand all the ways XML Sigs might 
> interact with OWL.
> In any case I hope such usage would be straightforward.
> 
> Jonathan
> 
> 

-----BEGIN PGP SIGNATURE-----
Version: PGP 6.5.8

iQA/AwUBPH0zvhdTablCCzU/EQL3DgCfbW1hFuBZa4lOfI9itqiGYPVTTRwAniDX
r0cKDsywL/FYo6QCw3rkeiML
=D0dC
-----END PGP SIGNATURE-----

Received on Wednesday, 27 February 2002 14:30:18 UTC