RE: RDF calendar schemas

Libby,

A couple of your last points/questions I think are already or could already
be addressed within the context of iCal.

iCal has a concept of Parent/Child relationships - this might be a way to
define the relationship between the conference and the seminars at the
conference for example.

There is also a RELATED_TO, and a SIBLING property for lateral relationships
(SIBLING is within the same tree, RELATEDTO is across trees).

Time related databases tend to store dates in some standardized way (usually
measuring increments from some fixed point - i.e. in Unix seconds since the
Unix epoch) - these are then converted back to
Day/Month/Year/Hour/Min/Second/microseconds dependant on the system. At this
point there are functions to extract this information.  The issues arise in
that there are some NON-transformable elements.

	i.e. it is NOT the case that ONE DAY == 24 hours (twice a year in areas
with Daylight Savings there are days that break this), nor can you state
that ONE MONTH == N Days - it depends on the Year and the Month in question
(and further, you can't equate ONE MONTH to hours, minutes, or seconds)

	Storing each field - Year, Month, Day, Hour, Minute, Timezone as separate
fields may lead to less connection between these elements - the time really
is ALL of the above working together (i.e. there are some times that simply
don't exist at all - just saying 2:30 AM is not enough to know this, you
also have to know the Year, Month, Day, and Timezone to determine if the
time is or not (see Daylight Savings)

There are many defined schemas for storing time, I do NOT suggest that for
this effort we reinvent the wheel, the issues involved are VERY complex.

Shannon

Shannon J. Clark
CEO - JigZaw, Inc
1.800.4.JIGZAW (454.4929)
shannon@jigzaw.com
www.jigzaw.com

-----Original Message-----
From: www-rdf-calendar-request@w3.org
[mailto:www-rdf-calendar-request@w3.org]On Behalf Of Libby Miller
Sent: Wednesday, May 30, 2001 5:55 AM
To: Greg FitzPatrick
Cc: 'Libby Miller'; www-rdf-calendar
Subject: RE: RDF calendar schemas


>
> That is great - I would like to know your opinion of Jonas's work
>
> http://lists.w3.org/Archives/Public/www-rdf-interest/2000Aug/0109.html
>
> How much of this do you find usable?

Some background: this url points to an RDF representation of Skical by
Jonas Liljegren. Skical [1] includes a representation system for public
events such as concerts, sports events. It builds on iCalendar and
extends it with information needed for public events such as
information about disabled access, availability of alcohol
etc. Jonas' work on Skical is highly relevant to work turning icalendar
into RDF.

I've described a strawman calendar schema [2] in order to have something
to play with which looks a lot like Jonas's description of Skical in
RDF, except for the dates. My schema uses rdf:label, which is
wrong. Otherwise, I'd agree that the structure of the RDF representation
can be essentially flat. Jonas' work is a good place to start. However,
in trying to query and use data like this, I have some issues I'd like
to bring up, concerning dates, repeated events and sub-events.

Dates

Should we be using XML schema datatypes to define dates? Especially
given the difficulties with the XSD namespace in RDF? In addition
querying anything which requires regexes or just partial string matching
as jonas's schema does within any of the current RDF apis isn't
possible, although query languages for RDF may well have
these capacities. This affects the form which dates can take.
Ideally we would want to have the option of separating out hour,
minute, date, month, year, into separate nodes as I started to do.

Repeated events

Repeated events are the most problematic area of calendaring. I
haven't looked into representating repeated events in RDF at all, so
I can't evaluate what Jonas has done yet.

sub events

Finally, I found it very useful in modelling the XMLEurope data to have
the notion of one event being contained within another. For example,
a track is a kind of event which might be spread over one or more days
at a conference, which contains various presentations, also events. Or
is a track is too broad for you, a session might contain more than one
event, and may plausibly be considered an event in itself. The
vaclendar is a container for events as I see it. We'd still need an RDF
property to indicate somehow that there was more data about the event
available (maybe rdf:seealso would do?).

Does this make sense?

cheers

Libby

[1] skical: http://ski.finns.nu/
[2] strawman RDF calendar
schema: http://ilrt.org/discovery/2001/05/ical/index.rdf

sample instance data
http://swordfish.rdfweb.org/discovery/2001/05/xmlecal/track1.rdf

XML Europe demo:
http://swordfish.rdfweb.org/discovery/2001/05/xmlecal/


>
>
> You have both the Dawson and Reddy DTD's as reference I hope?
>
>
>
> Greg
>
>
>

Received on Wednesday, 30 May 2001 12:57:24 UTC