Subject: Re: Data Element Naming Standards
Philip Coombs (pcoombs@wln.com)
Date: Mon, 28 Sep 1998 11:35:50 -0700
Message-Id: <199809281835.LAA13953@rs6a.wln.com> From: "Philip Coombs" <pcoombs@wln.com> To: <gils@cni.org> Subject: Re: Data Element Naming Standards Date: Mon, 28 Sep 1998 11:35:50 -0700
On Sat, Sep 26, 1998, Eliot Christian <echristi@usgs.gov> wrote:
>
> Folks interested in better mechanisms for metadata interoperability
> may want to check out the attached paper by Judy Newton of NIST
> <jnewton@nist.gov>. It addresses levels of abstraction in the context
> of data element names. I find the paper clear, concise, and short.
Eliot's message prompts further discussion on attribute naming.
It is time we consider the international aspects of data element naming
and allow for data element concept mapping / crosswalking across choice
of terminology, languages and character sets.
I propose the addition of a third component of indexing. This will be
the ISO equivalency of the data element concept (e.g., Title, Author,
date-of-publication) which would be described in arabic numbers. An
example:
First position: (Attribute name in local language / terminology)
Second Position: (Attribute Number)
Third Position: (Attribute value assigned by indexer / cataloger to the
specific object)
This would appear as an additional value in the attribute-value pair:
For instance:
Title <separator> 00004 <separator> =<separator> "The Name of the book"
or
Author <separator> 000017 <separator> = <separator> "A. Writer"
This arrangement could be used where simple attribute-value pairs are
used today.
Using naming conventions, each attribute chosen by an organization would
be associated with the ISO data element concept for interoperable
searching. Here is an example that could be used within HTML metatags:
<META Name = "Title" AttributeID = "00004" Content = "The name of this
book">
or in XML:
<Locator>
<Title>
<AttributeID>
00004
</AttributeID>
The name of this book"
</Title>
</Locator>
By using this convention, we could get beyond what terms we assigned for
common, conceptually-equivalent data elements. Also, associating the
attribute set heritage of the term would become moot: "DC.Title" and
"GILS.Title".
Demonstrating this concept, in French:
(Assuming the book is written in English)
<META Name = "Titre" AttributeID = "00004" Content = "The name of this
book">
The AttributeID maps meaning across languages and choice of terminology
and allows expanded use of the universal ISO character set.
Obviously a global attribute map would not hold up for all variations of
qualifier terms. At some point we must say that we can't recognize all
nuances of conceptual distinctions. When kept at a higher, common level
of qualifier (e.g., "Publisher" versus "Distributor", it should work
well.
I have read about proposals to build some global gateway for all
inquiries to go through and be translated. This would be unworkable.
By contrast, formatting quiries to recognize the AttributeID would place
the responsibility for concept mapping at the initiating / client site.
I welcome comments from the GILs listserv.
Philip Coombs
Washington State
360.902.5877
pcoombs@wln.com
This archive was generated by hypermail 2a16 : Tue Mar 23 1999 - 03:55:44 EST