OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

regrep-semantic message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]


Subject: regrep-owl-profile-1.0-draft 2


Dear Colleagues,

regrep-owl-profile-1.0-draft 2 is uploaded to the SC page.
In the following I will explain how the comments from Evan Wallace
and Farrukh Najmi are handled.

Farrukh Najmi wrote:

> The document is very well thought-out and written. The "OWL Overview" chapter is specially good as it conveys the key knowledge without assuming prior OWL or Ontology background. The document needs to specify Catloging/Validations/Discovery and other chapters similar to "ebXML Registry Profile for Web Services".


Cataloging and Discovery chapters are added.
We think that the Validation chapter is not relevant for this Profile.

> Line 5:
> Version 0.1 Draft 1 => Version 1.0 Draft 1


Corrected.

> Line 8:
>
> //Following comment is based on my undersatnding of OASIS doc naming conventions and how regrep specs
> //were identified. Things may have changed since.
> Jan16-2006regrep-owl-profile-1.0 => regrep-owl-profile-1.0-draft1
>
Corrected.

> Line 17:
>
> "This document defines the ebXML Registry profile for enhancing ebXML Registry with OWL
> semantics to make it OWL aware."
>
> => How about description below?
>
> "This document defines the ebXML Registry profile for publishing, management, discovery and reuse of OWL Ontologies"


In our opinion, this sentence may be misleading since we are mapping OWL ontologies
to ebXML; not storing them in their own syntax.

> Line 109-110:
>
> "and the predefined
> Association Types among RegistryObjects"
>
> => (drop pre-defined since assoc types are extensible)
>
> "and the
> Association Types among RegistryObjects"
>
Corrected.

> Line 111-112:
>
> "RegistryObjects can be associated with ClassificationNodes through
> Classification instances"
>
> =>
>
> "RegistryObjects can be classified using instances of Classification, ClassificationScheme and ClassificationNodes."
>
Corrected.

> Line 122-123:
>
> "without causing any changes in the ebXML Registry architecture
> specification"
>
> => (also do a global search / replace: "ebXML Registry architecture specification" => "core ebXML Registry architecture specification, [ebRIM], [ebRS]"
>
Corrected.

> "without causing any changes in the core ebXML Registry specifications, [ebRIM], [ebRS]"
>
Corrected.

> Line 133:
>
> "which can be used formalizing knowledge" => "which can be used *in* formalizing knowledge"
>
Corrected.

> Line 137-138:
>
> "Yet, OWL reasoners can not directly run on the ebXML registry
> because all the registry information is stored in OWL syntax."
>
> => (Above is note clear. Possible correction below:)
>
> "Yet, OWL reasoners can not directly run on the ebXML registry
> because all the registry information is *NOT* stored in OWL syntax."
>
Corrected.

> Line 185-186:
>
> "Within the scope of this document, only OWL Lite constructs are considered and in the rest of the
> document, OWL is used to mean OWL Lite unless otherwise stated."
>
> Do you think we want to follow above with:
>
> "A futrure version of this specification MAY support OWL DL"
>
ebXML OWL Lite Profile can very easily be extended to OWL DL
but since we will not be able to use this extra semantics in reasoning
I am not sure whether it worths doing it.

> Line 187-189:
>
> "In OWL, properties can have
> multiple domains and multiple ranges. Multiple domain (range) expressions restrict the domain (range) of
> a property to the intersection of the class expressions."
>
> Suggest having a reference to domain/range definition for those that are not familiar with these terms.
>
Evan Wallace suggested the following related with this item:

"I was actually surprised to see the above comment in this document as it is a fairly esoteric point.  It may be important to how the Profile was implemented,
but if not, I would suggest removing it entirely."

We agree with Evan and removed it completely.

> Section 2.1-2.9
>
> -Suggest replacing bullets with tables where a column is added with title "Description" that gives a brief one line description of the concept described by the bullet currently.
>
Evan had the following comment on this
"The OWL Overview chapter of the the Profile document seems to be mostly
excerpted from the OWL Overview [1] document that is part of the OWL specification.
Unless you have special permission to do this from W3C, this section will probably
have to be rewritten for the Profile."
We agree with Evan in the sense that the reference [McGuinness, Harmelen]
provides all the descriptions and examples. We thought it would not be nice to copy them.
However as long as they are properly referenced, I do not think that W3C
permission is needed just giving a list of OWL features.

> -Suggest
>
> Chapter 3:
>
> For now lets keep this chapter but suggest eventually removing this in later drafts and instead reference the same named chapter in a document being develoiped by Ivan Bedini called "ebXML Registry Deployment Profile Guide". We are moving away from duplicating this chapter in every profile.
>
> Line 465...:
>
> "It is important to note that although the mapping described in this section is complex, this complexity is
> hidden from the ebXML registry user because the needed stored procedures MUST already be available
> in the Registry as described in this chapter. As this profile aims to enhance ebXML registry semantics
> without causing any changes in the ebXML Registry architecture specification, the stored procedures
> proposed in this specification SHOULD be submitted to the ebXML Registry by using the Stored Query
> API of [ebRS]. It should be noted that arbitrary submission of the stored procedures defined in this profile
> would result in duplicate entries. Therefore, it would be more efficient to classify these stored procedures
> under a ClassificationScheme (e.g. urn:oasis:names:tc:ebxml-regrep:SemanticStoredProcedures)."
>
> I believe that the stored procedures are simply there to support the stored queries. The mapping is really enforced by the OWL Cataloging Service which need to be described in a chapter on Cataloging Profile. See example in "ebXML Registry Profile for Web Services"
>
Corrected.

> Line 522:
>
> regrep:AssociationType:Property =>  regrep:AssociationType:HasProperty
>
> This is more consistent with existing assoc types (+) but does not preserve the OWL element name exactly (-).
> If we do above change then we may want to suggest adding above to canonical AssociationType scheme as it is fairly generic.
>
Evan had the following comment on this:
"
I would strongly advise against this change. Properties are first

class elements in RDFS and OWL. What is it that "has" the Property above - an ontology?  If at all possible, please use the OWL, RDFS and RDF terms verbatim."

We agree with him.

> Line 538:
>
> "A new ebXML RIM Association Type called Property"
>
> =>
>
> "A new ebXML RIM Association Type called SubPropertyOf"
>
> Line 540...


Corrected.

>
> "This requires a recursion mechanism in SQL queries.
> The freebXML implementation allows various relational database products such as Oracle, PostgreSQL
> and MS SQL Server 2005 to be used as the database. These products have different support for
> recursion mechanism in SQL Queries. The following stored procedure is for retrieving all super properties
> of a given property instance (Association instance in ebXML terminology) recursively in a property
> hierarchy (hierarchy of Association Types) in freebXML Registry implementations using MS SQL Server
> 2005 as the database:"
>
> Is recursion mechanism in SQL queries. part os standard SQL? I suspect not. Also stored procedures are notorious
> for being db specific as you point out above. Suggest indicating the stored procedure as  Non-Normative implementation detail.
>
Corrected.

> Line 613:
>
> "OWL-S Profile"
>
> => Suggest using namespace prefix and have a table of all namespace prefixes somewhere in Intro to OWL chapter
>
> "owls:Profile"


Here is a comment from Evan on this:
"References to OWL-S would not belong in an OWL chapter since OWL-S is not

part of the OWL language, but rather a source of OWL ontologies used in
examples in the Profile document."

We agree with him.

We look forward to hearing your comments.

With best regards,

Asuman

-- 
____________________________________________________________________________
Professor Asuman Dogac   email: asuman@srdc.metu.edu.tr
WWW: http://www.srdc.metu.edu.tr/~asuman/
Director                 Phone: +90 (312) 210 5598, or
Software R&D Center             +90 (312) 210 2076
Department of Computer Eng.       Fax: +90 (312) 210 1004
Middle East Technical University       +90 (312) 210 1259
06531 Ankara Turkey                    skype: adogac


[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]