Title: Report from public comment period on DCMI Abstract Model
Creator: Pete Johnston
Identifier: http://dublincore.org/documents/2007/04/02/abstract-model/comments-received.html
Date: 2007-06-04
Period: 2 April - 30 April 2007
This document summarises the principal comments received during
the public comment period on the DCMI Abstract Model (Version
2007-04-02) Proposed Recommendation [1], a revised version of
the previous Proposed Recommendation (Version 2007-02-05) [2].
[1] http://dublincore.org/documents/2007/04/02/abstract-model/
[2] http://dublincore.org/documents/2005/02/05/abstract-model/
Proposals for moving forward are presented in the sections below.
OWL-DL, datatype properties and object properties
OWL-DL will require properties to be declared as either
datatype properties & object properties. If both literal
and non-literal values are required, then may need two
parallel properties.
This is being taken into account by the Usage Board in
their consideration of the ranges of DCMI properties.
Proposals: No change required.
References:
-- Comment: Ivan Herman, 2007-04-27
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0704&L=dc-architecture&P=1308
-- Reply: Pete Johnston, 2007-05-20
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0705&L=dc-architecture&P=424
Specification of which value string is intended for display
If statement has a non-literal value surrogate with
multiple value strings, how does the application decide which
string to display?
DCAM does not provide this but can make use of the description
of a value to differentiate "categories" of label e.g. using
SKOS preferred/alternative label properties.
Proposals: No change required.
References:
-- Comment: Douglas Campbell, 2007-05-01
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0705&L=dc-architecture&P=56
-- Reply: Pete Johnston, 2007-05-20
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0705&L=dc-architecture&P=700
Definition of literal
Provide fuller definition of literal rather than just
referencing RDF Concepts.
Proposals: Expand description/definition of literal in
DCAM, based on RDF C&AS, XML Schema.
References:
-- Comment: Douglas Campbell, 2007-05-01
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0705&L=dc-architecture&P=56
-- Reply: Pete Johnston, 2007-05-20
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0705&L=dc-architecture&P=700
Requirement to use URI to reference VES & SES
The DCAM requires that properties, VES & SES be referred
to in statements within DC metadata description sets
using URIs.
Proposals: No change required.
References:
-- Comment: Douglas Campbell, 2007-05-01
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0705&L=dc-architecture&P=56
-- Reply: Pete Johnston, 2007-05-20
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0705&L=dc-architecture&P=700
Distinction between Literal Value Surrogate and Non-Literal Value Surrogate
Distinction is not based on number of value strings
present in statement, but on type of value itself.
Proposals: No change required.
References:
-- Comment: Douglas Campbell, 2007-05-01
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0705&L=dc-architecture&P=56
-- Reply: Pete Johnston, 2007-05-20
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0705&L=dc-architecture&P=700
Language and Typed Value Strings/XML Literals
Typed literals cannot carry language tags.
Proposals: No change required (though maybe need to emphasise that
typed value strings includes XML Literals?).
References:
-- Comment: Douglas Campbell, 2007-05-01
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0705&L=dc-architecture&P=56
-- Reply: Pete Johnston, 2007-05-20
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0705&L=dc-architecture&P=700
dcam:memberOf equivalent in RDF
RDF has no "built-in" equivalent of VES, so no built-in equivalent of dcam:memberOf.
Proposals: No change required.
References:
-- Comment: Douglas Campbell, 2007-05-01
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0705&L=dc-architecture&P=56
-- Reply: Pete Johnston, 2007-05-20
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0705&L=dc-architecture&P=700
Definition of "term"
Definition of "term" should be part of bullet list in section 2.3.
Proposals: Add bullet defining term near start of list.
References:
-- Comment: Douglas Campbell, 2007-05-01
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0705&L=dc-architecture&P=56
-- Reply: Pete Johnston, 2007-05-20
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0705&L=dc-architecture&P=700
Set of resources (VES) v Set of literals (SES)
Not always clear when to use VES v when to use
SES. A modelling choice. Depends on requirements of
application. Usage Board choices for DCMI terms based on
content of existing specifications.
Proposals: No change required.
References:
-- Comment: Douglas Campbell, 2007-05-01
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0705&L=dc-architecture&P=56
-- Reply: Pete Johnston, 2007-05-20
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0705&L=dc-architecture&P=700
Description of "subproperty"
Bullet on subproperty in 2.3 should use the phrasing
"...and the property is part of a property/value pair"
(rather than "...and a described resource is related to
a value by that property")
Proposals: Change phrasing of sixth bullet in section 2.3.
References: Offlist comment by Pete Johnston
Error in 2.4
Erroneous reference to "syntax encoding string"
Proposals: Change "syntax encoding string" to "syntax encoding scheme".
References: Offlist comment by Pete Johnston
Abstract Syntax Required
Alistair Miles re-emphasised the points made in the
previous comment period: the need to delineate an abstract
syntax, to describe how to merge DC description sets,
and to describe inference rules. He also suggested the
need for use cases to specify the requirements that the
DCAM is intended to meet.
Proposals: In short term, move forward with current
version; in longer term, gather requirements using current
DCAM as input, and consider revision/restructuring along
lines suggested.
References: Discussion during telecon, 2007-05-17
Definition of "vocabulary"
DCAM uses "vocabulary" to refer to set of "terms"; RDF
Semantics uses "vocabulary" to refer to set of names
(URIs/literals).
Proposals: Current DCAM use is consistent across DCAM
and Namespace Policy document; usage in W3C docs does
vary. Retain current use in DCAM.
References: Discussion during telecon, 2007-05-17