iCalendar in RDF - first draft

libby.miller@bristol.ac.uk 2001-06-12

examples

Full draft RDF schema

Just the classes

Just the datatype classes (not dates) [gif]
Just the date-related classes [gif]
Just the calendar object-related clases [gif]

Just the properties

Just the properties with CAL-ADDRESS as their domain [gif]
Just the properties with TEXT as their domain [gif]
Just the properties with DATE as their domain [gif]
Just the properties with DATE as their range [gif]
Just the properties with URI as their domain [gif]
Just the properties with VCALENDAR as their domain [gif]
Just the properties with REC-CAL-COMPONENT as their domain [gif]
Just the properties with VJOURNAL as their domain [gif]
Just the properties with VALARM as their domain [gif]
Just the properties with VDOTO as their domain [gif]
Just the properties with VFREEBUSY as their domain [gif]
Just the properties with VTIMEZONE as their domain [gif]
Just the properties with VEVENT as their domain [gif]
Just the properties with RECUR as their domain [gif]

Notes

• Cal-address is an ID for a person or thing that may or may not use the calendar. In the RFC 2445 it is treated as an object though, not a mail address or similar, although the RFC 2445 sometimes also uses its ID as a mail address.
• What is the relationship of rec-cal-component and cal-component and the individual components?
  
  i.e. There are six calendar components: VEVENT, VTODO, VJOURNAL, VFREEBUSY, VALARM, VTIMEZONE. Of these, VEVENT, VTODO, VJOURNAL, VFREEBUSY are fairly similar, so I've made them subclasses of a new class called CAL-COMPONENT. However, of these, VFREEBUSY cannot be recurring. So I created a new class called REC-CAL-COMPONENT which has subclasses VEVENT, VTODO, VJOURNAL, and which indicates that event is a recurring one (not that it _may_ recur). In addition, certain properties can have as domain or range almost any combinination of the six calendar components. Since there can be multiple domains, this is ok, but you can't have multiple ranges in RDF.
• What is the relationship of date and time to date-time and each other? Aaron suggested:
  
  "create two types (Date-Time and Date) and then create a new class (TimeEntry) that contains both of them as subclasses. Then point to TimeEntry"
  
  I've created a DATE object which is the superclass of DATE-TIME. I don't know what to do with TIME. Maybe DATE-TIME should be the superclass fo DATE and TIME. I'm just not convinced of which is the more general.
• Should we use literals or text where alternative literals are specified (or are specified in RDFs)? For most cases it seems correct to make TEXT a subclass or resource not literal, because it requires langauge tagging and sometimes other information be attatched to the node. But if this is how we do it, where the value of a TEXT is explicilty given (e.g. PUBLIC /PRIVATE/CONFDENTIAL for CLASSIFICATION) then that might indicate that we ought to have a literal instead. maybe.

Links

RDf calendar home page

Calendar links

A quick look at iCalendar

iCalendar in UML

Using XML Schema Datatypes in RDF and DAML+OIL

DAML-S Semantic Markup for Web Services (PDF)

DanCs Palm datebook schema in RDF/XML

DanCs n3 Palm datebook schema

ABC A Logical Model for Metadata Interoperability

Proposed RSS event module

Greg Fitzpatrick/Jonas Liljegren notes on XML/RDF calendaring