- to use a general purpose calendar vocabulary,
or a specific one for the semantics the device understands? The code currently uses a palm-specific vocabulary; ontology mapping is full of subtleties that are to be handled above its service interface. For example, suppose the code used the RdfCalendar encoding of iCalendar a request came thru the service interface to insert an event in America/New_York time. The data store doesn't grok timezones. Should it throw an exception? throw away the information? Meanwhile, it's easy enough to map palm datebook semantics to icalendar semantics above the service layer with rules such as pd2ical.n3, and with extra information about what timezones the palm device will be in, it's not too hard to map the other way.
is maintained in Notation3 and converted to RDF/XML now, but it was originally available in HTML and RDF. Is negotiating between HTML and RDF appropriate?
- modelling database records and C-style structures: the palm schema is isomorphic
to an SQL store; the schema should encode the "and that's all" property of records and the service interface should support update as well as insert.
see also: SemanticWebModellingIssues