Thanks Alex. Just a few points:
1) I wouldn't advocate putting the SICI into dc:date. The SICI is a unique
issue and article identifier, and as such should go into dc:identifier. A
"SICI-savvy" interrogator would be able to divine that the issue was, for
example, the Spring issue from the Chronology piece of the SICI.
2) In addition to using the SICI, you could use the "citation string", as
recommended by the dc-citation Working Group (which also recommended that
this should be indicated by a new qualifier, dc:identifier.citation). This
could also contain the explicit term "Spring" in its Chronology piece, but
this is not specifically designed to be parsed since for that you would
tread into the murky depths of Dublin Core Structured Values.
3) And in addition to (1) and (2), you could put "Spring 2000" into the
dc:date field, just as a regular text string.
This is all a bit belt and braces, but since DC is designed for resource
discovery, not for database sorting, repetition of information in different
fields may be appropriate. In raw DC, nothing's mandatory anyway. (As Ann
Apps has suggested, if more rigour is needed, this is best dealt with by
the use of application profiles - *then* you can add structure and parse
merrily away.)
4) There's always room in DC for human-readable data values. It's not
optimised for machine-to-machine communication.
Regards
Cliff (who wishes he could come up with a quote as good as your
"interoperability standard" definition)
Alex Satrapa <[log in to unmask]> on 21/12/2000 23:49:52
To: [log in to unmask]
cc: [log in to unmask]
Subject: Re: ISO date
At 09:20 +0000 2000-12-21, [log in to unmask] wrote:
>It seems to me this whole debate arose from a wish to use dc:date to
>describe this information. ...
[snip]
>... I'd say you had four (not
>mutually exclusive) choices:
>
>ii) You could use a SICI in dc:identifier, where the Chronology piece
would
>indicate Spring by "21".
So rather than using the ISO-8601 scheme with non-ISO-8601 dates, use
the SICI scheme with SICI codes? Sounds like a good idea to me :)
>iii) You could follow the recommendations of the dc-citation Working Group
>and put the following into dc:identifier for the issue - "Journal of
>Metaphysical Metadata (Spring 2000), Volume 1, Issue 1". (See the
>discussion list archives at
>http://www.jiscmail.ac.uk/lists/dc-citation.html.)
This is good for human-readable content, but for machine-readable
versions, my choice would be to have a repeated dc:date using the
SICI code. An alternative is to attempt to parse the string, in the
hope that the people encoding it are following the same "standard"
that you are, and that the encoding scheme prevents ambiguity in the
encoding.
I wonder if there's room in a "Dublin Core Query Language" to specify
"I understand these schemes for these fields"?
>iv) You could mention that it's the Spring issue in dc:description.
Once again, great for getting the message across to humans, but not
necessarily good for machine-readability (required for things like
sorting or searching).
Regards
Alex Satrapa
FWIW: "Interoperability Standard: n. A delicate balance between the
laziness of database and human interface programmers, the hubris of
the data structures experts, and the impatience of stakeholders and
sponsors."
|