----------------------------------------------------------------------
Date: Tue, 21 Mar 2006 11:04:37 +0000
From: Pete Johnston
Subject: Re: Updated Wikipedia article
To: DC-USAGE@JISCMAIL.AC.UK
Diane I. Hillmann wrote:
> Folks:
>
> I went in and changed the page: http://en.wikipedia.org/wiki/Dublin_Core
> using stuff from "Using Dublin Core" primarily.
>
> You're welcome to see what you think and do some editing yourself, if
> you've a mind to do so. I won't take it personally, I promise!
I haven't edited the document on Wikipedia (yet), and I don't really
want to do so unilaterally, but (and this really applies to other
introductory resources like the "Using Dublin Core" document and like
the conference tutorials) I think that the role of the DCMI Abstract
Model as providing the conceptual framework for DC should be presented
more "up front" e.g. there should be a short/simplified summary
description of the DCAM in the introductory section, or as a separate
section following the intro and before the discussion of Simple DC and
Qualified DC. Mentioning the DCAM _only_ as a tool for comparing
different syntaxes is only telling half the story (IMHO) - though I
think the DCAM doc itself might be slightly more "bullish" in its own
intro paragraph! ;-).
Now sure, I appreciate that that might look like a case of "historical
revisionism" to the casual observer who knows "the 15 elements came
first" (or indeed only knows the 15 elements), but I do think we need to
shift firmly towards putting the DCAM at the centre of our explanations
of "what DC is". (If people want a history of the evolution of DC, and
how DCMI got from "the 15 elements" to "the qualifiers" to the
grammatical principles to the DCAM, OK, that's fine, but that's a
different document.)
I recognise this probably goes against the way we've tended to introduce
DC, but I'd go as far as saying that it is confusing/unhelpful to start
talking about "elements" without first describing the DCAM, at least in
some way - maybe not every fine detail, but the fundamental points about
making statements that assert relationships between resources and
values. Without such "contextual" information, it just begs the question
of what an "element" is. It makes a "leap of faith" that readers already
share a common understanding of what an element is, but (as we've found
out somewhat painfully over the last few years), that is not the case:
the term "element" is used to refer to different things in different
contexts and readers draw their own (different, incompatible)
conclusions ("Ah, they're talking about XML elements", "Ah, so they're
referring to things like LOM elements", "Ah, they mean attribute-value
pairs" etc etc etc).
I think the account of "Simple Dublin Core" also blurs the distinction
between the DCMES as a set of terms, each of which may be deployed in
many different "DC application profiles" with different constraints on
their usage in a description set, and "Simple Dublin Core" as one such
DCAP with one particular set of constraints. And in the account of
"Qualified DC" I'm not sure the word "value" is being used in the way it
is used in the DCAM. I think phrases like "the value may still be useful
to a human reader" suggest that the reference is to (what the DCAM
calls) "value strings".
----------------------------------------------------------------------
Date: Tue, 21 Mar 2006 11:55:30 -0000
From: Andy Powell
Subject: Re: Updated Wikipedia article
To: DC-USAGE@JISCMAIL.AC.UK
Yes, I agree with this... though it's certainly not easy to come up with
the one paragraph summary of what DC is. That's one of our problems!
But talking about 'elements', 'simple' and 'qualified' doesn't help
much!
Here's my stab at a summary...
--- cut ---
Dublin Core (DC) is a metadata standard for describing a wide range of
digital, physical and conceptual resources (i.e. just about anything!).
A DC description is made up of a set of statements, each of which
comprises a property/value pair. Typically, the described resource is
identified using its URI and the value is either identified using its
URI or represented using a simple string (the 'value string'). In many
cases, multiple descriptions are combined in order to build up richer
descriptions ('description sets') about related groups of resources.
For example, in describing a digitised painting, it may also be
appropriate to describe the original painting from which the
digitisation was made and the original artist, thus creating a
description set of three related descriptions. Statements may be
refined by indicating the language of the value string, any data-type
('syntax encoding scheme') to which the value string conforms or the
class ('vocabulary encoding scheme') from which the value is taken.
Properties, syntax encoding schemes, vocabulary encoding schemes and
concepts in controlled vocabularies are known as 'terms'. All terms in
DC metadata are assigned URIs, and schema languages are used to indicate
the relationships between them.
The features of the DC metadata standard are fully described in the
Dublin Core Abstract Model (DCAM).
The Dublin Core Metadata Initiative (DCMI), the body that facilitates
the community development of the DC metadata standard, provides a core
set of about 80 properties, encoding schemes and controlled vocabularies
from which descriptions can be constructed, but encourages communities
to create additional terms as necessary, within the framework provided
by the DCAM.
Finally, the DCMI community has defined three encoding syntaxes that can
be used to encode DC metadata records for exchange between software
systems and services using XHTML, XML and RDF.
Historically, DC refered to properties as 'elements' and is perhaps best
known for the Dublin Core Metadata Element Set (DCMES) - a set of 15
elements, created originally to support the discovery of resources on
the Web.
--- cut ---
This, rightly(!), relegates DCMES to a footnote and doesn't even mention
simple and qualified! :-)
On the face of it, it may seem harder to grasp than the traditional
'element', DCMES, 'simple DC', 'qualified DC' approach - but I think
that is largely to do with where we come from. Personally, I think it
is much clearer - it emphasises what is important and ignores what can
safely be forgotten.
----------------------------------------------------------------------
Date: Tue, 21 Mar 2006 12:06:53 -0500
From: "Diane I. Hillmann"
Subject: Re: Updated Wikipedia article
To: DC-USAGE@JISCMAIL.AC.UK
Tom, responding to Pete and Andy:
>I agree that the Abstract Model should be at the center of
>explanations, and I like most of the text. I also think we
>need to revisit the explanation for "simple" versus "qualified"
>Dublin Core.
I agree that the Abstract Model should be better integrated into the
text, and have said more than once that I'd like to find a way to
integrate the AM better into "Using Dublin Core," but could use some
help with that. Because I used text from "Using Dublin Core" as the
basis for what I added to the article, it suffers from some of the
same problems.
>Two issues however:
>
>1) To start with statements, property-value pairs, and URIs is
> to jump in at the deep end. The notion of "core metadata
> properties" has been and should continue to be a key part
> of the message. Introducing "the Dublin Core" up-front also
> helps explain the funny name.
>
Yes, in my experience, the DCAM certainly is the deep end for most
people. I think it makes more sense to use Wikipedia to provide a
"soft" approach, dealing first with things people might have already
heard about DC, and making that clearer before getting into deeper
water.
> > The Dublin Core Metadata Initiative (DCMI), the body that facilitates
> > the community development of the DC metadata standard, provides a core
> > set of about 80 properties,
>
> I suspect that the notion of "core metadata properties" really
> is easier for most people to grasp than the notion of an
> abstract model. I would not want DCMI to lose that focus --
> 80 properties is no longer really a "core".
Agreed. Despite our mere 10 years of "legacy," we are seen by many in
an increasingly complex world as more approachable and
understandable--metadata for the masses, perhaps. We can't afford to
lose that advantage.
>2) "DC" and "DCMI"
>
> > The features of the DC metadata standard are fully described
> > in the Dublin Core Abstract Model (DCAM).
Usage aside, I think the sentence above denies the fact that the
explanation provided in the DCAM, while certainly full, is fairly
technical from the point of view of the metadata novice.
> This is not quite how I understand our current use of
> "DC" and "DCMI". To take two important examples, we
> currently have the "DCMI Abstract Model" and we talk
> about "Dublin Core Application Profiles" (but not "DC
> Application Profiles" or "DCMI Application Profiles").
> In a reasonably consistent way, we currently:
>
> -- use "Dublin Core" instead of "DC". The exception is when
> we use the names of Jiscmail lists as handles for working
> groups (e.g., "DC-Architecture"). In DCMI publications,
> we have I think avoided using the free-standing acronym
> "DC" to mean just "Dublin Core" for quite a long time.
> If we were to revive it, would we do so with the intent
> to refer to "the Dublin Core" (DC-15) or to designate
> the abstract model?
>
> -- use "DCMI" for things that are managed or maintained by
> DCMI as an organization -- e.g., "DCMI Usage Board".
> The DCMI Abstract Model is a model put forward and
> maintained by DCMI for metadata that uses DCMI terms,
> (whereas a Dublin Core Application Profile is in most
> cases not put forward or maintained by DCMI).
I *think* I followed most of this pattern, but may have blundered a bit ... ;-)
----------------------------------------------------------------------
Date: Wed, 22 Mar 2006 07:52:51 +0000
From: Pete Johnston
Cc: DC-USAGE@jiscmail.ac.uk
Subject: Re: Updated Wikipedia article
Quoting Thomas Baker :
> Two issues however:
>
> 1) To start with statements, property-value pairs, and URIs is
> to jump in at the deep end. The notion of "core metadata
> properties" has been and should continue to be a key part
> of the message. Introducing "the Dublin Core" up-front also
> helps explain the funny name.
>
> > The Dublin Core Metadata Initiative (DCMI), the body that facilitates
> > the community development of the DC metadata standard, provides a core
> > set of about 80 properties,
>
> I suspect that the notion of "core metadata properties" really
> is easier for most people to grasp than the notion of an
> abstract model. I would not want DCMI to lose that focus --
> 80 properties is no longer really a "core".
Unless we explain what we mean by "property", and show how properties
are used in DC metadata (to construct simple statements that assert
relationships between resources), then we are left with the same
problem with the word "property" that we have now with the word
"element".
Agreed, "property" maybe comes with slightly less "baggage" than
"element", but I've had experience of people picking up on my tendency
to use "property" rather than "element" and altering their usage
accordingly. But unless I explain why I'm saying property - what a
property is - then they think I'm using it simply as a synonym, and
they start using "property" for all the contexts in which they were
using "element", including ways which are inconsistent with the DCAM.
And - perhaps even worse - they start to apply the term "property" to
things defined in other meta-models - where those thinga are not
properties at all, and the "native" (to that meta-model) term "element"
is indeed the most appropriate term.
Whatever we call the fifteen things, we have to explain what they are
and how they are used.
----------------------------------------------------------------------
Date: Wed, 22 Mar 2006 13:34:49 -0000
From: Andy Powell
Subject: Re: Updated Wikipedia article
To: DC-USAGE@JISCMAIL.AC.UK
> I suspect that the notion of "core metadata properties" really
> is easier for most people to grasp than the notion of an
> abstract model. I would not want DCMI to lose that focus --
> 80 properties is no longer really a "core".
OK, could drop the specific ('80') and leave the general ('core set of
...')?
But main problem is with emphasis and order. It's difficult to say what
DCMI provides (essentially 'a core set of terms that can be used to
create descriptions of resources') without describing the nature of what
those terms are and how they fit together.
Perhaps in my suggested text,
The Dublin Core Metadata Initiative (DCMI), the body that facilitates
the community development of the DC metadata standard, provides a core
set of about 80 properties, encoding schemes and controlled vocabularies
from which descriptions can be constructed, but encourages communities
to create additional terms as necessary, within the framework provided
by the DCAM.
needs to be moved up front and changed to
The Dublin Core Metadata Initiative (DCMI), the body that facilitates
the community development of the Dublin Core metadata standard, provides
a flexible and extensible metadata framework and a core set of terms
from which descriptions can be constructed, but encourages communities
to create additional terms as necessary to meet their own descriptive
requirements.
Then go on to describe the framework (the DCAM) in more detail...
Does that help?
> 2) "DC" and "DCMI"
>
> > The features of the DC metadata standard are fully described
> > in the Dublin Core Abstract Model (DCAM).
>
> This is not quite how I understand our current use of
> "DC" and "DCMI".
Sorry. I was writing in a hurry so I didn't follow corporate
guidelines! :-)