David:
The problem is twofold, as I see it: if we indeed include an "indentifier",
we provide very little real illustration of what we're talking about, since
identifiers are most often extremely cryptic--useful for linking, but not
much else. Secondly, in doing this sort of thing, instead of sticking to a
purely textual solution, we are skating very close to the edge of (if not
going over into) qualification, which has made me somewhat uncomfortable
all along, and is why I suggested a couple of weeks ago that we back up
into a less structured solution for Relation.
The solution now in V5 was essentially the one suggested by Simon many
months ago, and everyone else has looked at it and made little if no
comment up til now. What would you have me do at this point? We need to
finish this process, that much is clear.
My "simplier" proposal, in May, looked like this:
*******
I'd like to suggest that we accept the proposal and simplify that
section to say:
[eliminate the list of types of relationships]
Guidelines for content creation:
Express as clearly as possible the relationship between the related
resource and the one being described.
Examples:
[Being described: Elton John's 1997 song Candle in the Wind]
Relation = Updated version of Elton John's 1976 song Candle in the
Wind
[Being described: the movie My Fair Lady]
Relation = Based on George Bernard Shaw's play Pygmalion
OR,
[Being described: Review of the movie Titanic]
<META NAME = "DC.Relation" CONTENT = "Titanic">
*******
Pretty much the only response I got was from Simon, who hated it, and
another person who agreed with him. So it stayed as is.
I would much prefer a minimalist solution for Relation in "Simple Dublin
Core" frankly, and leave the structured solution to "Qualified Dublin
Core," whenever and however that rolls around. Mandating an indentifier for
Relation (and Rights, for that matter) seems to me the imposition of a
structure where we have already decided we would not press for one.
Diane
>Dear Diane,
> According to the guide, which I agree with;
>
>"The recommended method for expressing a relationship in unqualified DC is:
>
>Title="the present resource"
>Relation="relationship-type [space] unique identifer for the related
>resource"
>
>
>But _none_ of the examples include a "unique identifier for the related
>resource"
>
>It doesn't seem to me to even illustrate what we think is the best
>practice, to say nothing of recommend it.
>
>David
>
>
>At 03:14 PM 6/9/98 -0400, Diane I. Hillmann wrote:
>>Folks:
>>
>>We're down to the wire. The "User Guide for Simple Dublin Core" is now in
>>final draft, and will go to Stu Weibel as a finished document on Friday
>>morning, barring any screams of protest.
>>
>>I have made a couple of itsy-bitsy changes, mostly fixing typos. Also, I
>>have changed the wording in the first paragraph where "Metadata" is
>>defined, in response to some late-breaking comments. The former version
>said:
>>
>>"Metadata is, simply stated, a description of an information resource. The
>>term "meta" derives from the Greek word for change; "metadata," then, is
>>data that documents the origins of, and/or tracks the change or use of,
>>data. Metadata may be used for a variety of purposes: to identify a
>>resource to meet a particular information need; to evaluate the quality or
>>fitness for use of such a resource; to track the characteristics of a
>>resource for subsequent maintenance or usage over time; and so on. ..."
>>
>>The new version says:
>>
>>"Metadata is, simply stated, a description of an information resource. The
>>term "meta" derives from the Greek word "denoting a nature of a higher
>order or
>>more fundamental kind," such as metalanguage, or metatheory. "Metadata,"
>>then, is data that sits above other data. Metadata may be used for a
>>variety of purposes: to identify a resource to meet a particular
>>information need; to evaluate the quality or fitness for use of such a
>>resource; to track the characteristics of a resource for subsequent
>>maintenance or usage over time; and so on. ..."
>>
>>
>>Anyone who wants to take potshots at my new wording, feel free. Do it
>>quick, though ... ;-)
>>
>>Diane
>>
>>P.S. Many thanks to all those of you who assisted with the work of
>>fashioning and revising this guide. Your comments have been instrumental
>>in making this document as useful as it can be. Please do stick around for
>>the first draft "Guide to Complex Dublin Core" coming soon (perhaps by this
>>fall) to a screen near you!
>>
>>
>>
>>
>David Bearman
>President
>Archives & Museum Informatics
>5501 Walnut St., Suite 203
>Pittsburgh, PA 15232
>tel. +1-412-683-9775; fax +1-412-683-7366
>http://www.archimuse.com
|