Re: ical and geo/location (was Re: A Japanese ...)

On Tue, 2003-11-11 at 08:01, Masahide Kanzaki wrote:
> Hi, I've been thinking about geo and iCal for a while.

Me too... ;)
http://www.w3.org/2000/10/swap/pim/travel.html

[...]
> If we want to add some information on NY Sheraton (not the event) such as
> url of the Hotel homepage or photo of the hotel, this can be extended as:
> 
> <Vevent>
>  <location rdf:parseType='Resource'>
>   <dc:title>New York, Sheraton Hotel</dc:title>
>   <geo:lat>40.442673</geo:lat>
>   <geo:long>-79.945815</geo:long>
>   <foaf:homepage rdf:resource="..."/>
>   <foaf:depiction rdf:resource="..."/>
>  </location>
>  ...
> </Vevent>
[...]

> Hmm.., but this requires the change of range of <location>, and "changing
> the range of properties could cause problems" as Morten noted... then, is
> it better to use <geo> of ical for this structured description (not a good
> fit name, but is not defined in the schema yet) ?
> 
> I prefer using <location> for structured description, if possible.

Well, our strategy so far is to do a fairly "dumb" mapping from
properties in the ical namespace...
  http://www.w3.org/2002/12/cal/ical#location

to the IETF proposed standard semantics:

[[[
4.8.1.7 Location

   Property Name: LOCATION

   Purpose: The property defines the intended venue for the activity
   defined by a calendar component.

   Value Type: TEXT
]]]
  -- http://www.ietf.org/rfc/rfc2445.txt

While I agree this is a modelling error (cf
http://esw.w3.org/topic/ThingsVersusTheirNames) I want us
to deal with modelling errors by, for example, using other
properties and mapping them to ical:location, rather than
teaching our ical/RDF tools to be smart enough to "fix"
the modelling error at mapping time.

To relate an event to a place (rather than a place's name)
I use cyc:eventOccursAt. e.g.

[[[
@prefix cn: <http://opencyc.sourceforge.net/daml/cyc.daml#>

...

_:webConf a :Conference;
  :homePage <http://www2002.org/>;
  cn:eventOccursAt [
    a :Hotel; :name "Sheraton Waikiki";
    usps:deliveryAddress "2255 Kalakaua Avenue";
    usps:zipCode "96815";

...
]]] -- http://www.w3.org/2000/10/swap/test/pathCross.n3


So a mapping rule might look like:

{ ?E cyc:eventOccursAt [ cyc:nameString ?NAME ] }
  log:means { ?E ical:location ?NAME }.


The cyc ontology has great documentation...
Darn it; the link from
  http://www.w3.org/2000/10/swap/pim/travel.html
to
  http://www.cyc.com/cyc-2-1/vocab/transportation-vocab.html
has gone bad... ah; here it is:
  http://www.cyc.com/cycdoc/vocab/transportation-vocab.html
(sigh; why 404 rather than a simple redirect?)
There:
  http://www.cyc.com/cycdoc/vocab/actor-vocab.html#eventOccursAt


I'm mostly happy to use cyc, but sometimes I think about
switching to SUMO for licensing reasons; they have an RDF
version...
  http://reliant.teknowledge.com/DAML/SUMO.daml
I've talked with them about splitting it into about a dozen
manageable chunks. Anyway...

Their wordnet-based browser is great...
http://ontology.teknowledge.com:8080/rsigma/SKB.jsp?req=SC&name=Entity&skb=SUMO

There's usually an analog there... ah... perhaps sumo:located
http://ontology.teknowledge.com:8080/rsigma/SKB.jsp?req=SC&skb=SUMO&id=100




> 
> cheers,
-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Tuesday, 27 January 2004 17:57:27 UTC