innovation in metadata design, implementation & best practices

Title: Problems with the definition of dc:date
Identifier: http://dublincore.org/usage/meetings/2004/10/ISSUES/definition-date/
See also: http://dublincore.org/usage/meetings/2004/10/ISSUES/
Created: 2004-09-14
Agenda frozen: 2004-10-02 07:25, Saturday
Archived: 2004-11-10
Maintainer: Tom Baker
Note: If any of the links below are broken, please refer to 
                   the meeting packet
                   (http://dublincore.org/usage/meetings/2004/10/Meeting-packet.pdf) 
                   for copies of the key documents discussed at the meeting.

Andy stated the problem on 2004-07-27 on the DC-USAGE list:

    This issue has popped up twice on the lists recently.
    dc:date is defined as

      A date of an event in the lifecycle of the resource.

    which superficially looks like the intention was a single
    date rather than a date range.

    But I tend to agree with what Charles says here.
    Any single date is really a shorthand for a time-range
    (00.00 to 23.59 on the given day)... so what I suspect
    we really meant by dc:date was

      A date/time (or date/time range) of an event in the lifecycle of the
      resource.

    Would it be worth trying to clarify this in our
    documentation? Or do people disagree with my analysis
    anyway?

    ---------- Forwarded message ----------
    Date: Mon, 26 Jul 2004 21:23:23 -0400 (EDT)
    From: Charles McCathieNevile <charles@w3.org>
    To: kurt.godden@gm.com
    Cc: www-rdf-interest@w3.org
    Subject: Re: Dublin Core 'available' date range
    Resent-Date: Mon, 26 Jul 2004 21:24:06 -0400 (EDT)
    Resent-From: www-rdf-interest@w3.org

    You could model its availability as an event with a start and end time using
    the RDF/iCal work - http://www.w3.org/2002/12/cal/

    Although it is hard to say if this is the same as what dublin core thinks a
    single date might be. If it doesn't, it probably should. Time being a
    continuum (at least as far as we measure it) any date or time represents a
    range...

Pete replied:

    (I was involved in one of the discussions Andy mentioned above [1])

    I was sort of willing to be persuaded by Charles McC-N's message to rdf-
    interest too. i.e. a date is always really a reference to a period of time.

    But.... ISO8601:2000 does make a distinction between

    (a) a date, defined as

    ====
    identification of a particular calendar day, expressed by some combination
    of the data elements calendar year, calendar month, calendar week, calendar
    day or day of the year
    ====

    (b) a time interval or period of time, defined as

    ====
    portion of time between two time points
    ====

    I was kind of surprised to read that a date is always a reference to a
    _day_ but that's what it says.... (it may be of reduced precision)

    If the intention is that a value of dc:date can be either a date, a
    combination of date and time, or a time interval/period of time -
    especially as DCMI has accepted ISO8601 as an encoding scheme - I think the
    terminology in DC definition probably should be consistent with ISO8601
    usage, and there's a case for amending the definition of dc:date.

    FWIW, I think that would still leave open the possibility that some
    refinements of dc:date could specify that their values are specifically
    dates (and not time intervals), if that was required.

    Incidentally, I notice that ISO8601 also supports the representation of
    time of day alone. But I'm assuming a time that is not associated with a
    date is not an acceptable value for dc:date? I'm not sure whether that
    merits some qualifying note in the (forthcoming!) description of ISO8601 as
    an encoding scheme.

    You can get a final(?) draft of ISO8601:2000 at [2]. But I guess you have
    to buy the approved version.

    Pete

    [1] http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0407&L=dc-collections&T=0&F=&S=&P=2734
    [2] http://lists.ebxml.org/archives/ebxml-core/200104/pdf00005.pdf