Re: WEBONT "HOMEWORK" (DUE DATE approaching!)

Like Dan C, our organisation has effectively been using
something like daml:uniqueProperty to talk about the organisation with
the homepage http://ilrt.org, or the person with the email address
libby.miller@bristol.ac.uk in RDF.

Dan Brickley has written an example scenario at

http://rdfweb.org/2001/01/design/smush.html

I think this technique also has implications for storing RDF, for
example, if you know that the node at the end of a daml:uniqueProperty
is a unique key in the database, you can index the data effectively.

Our work with RDF to date has not involved complex ontologies - we have
been using the experimental foaf RDF schema
(http://xmlns.com/foaf/0.1/) for useful things like mbox, name, knows,
currentProject and interest, of people, documents, images,
organisations.

I'm interested in the daml datatyping work, because of my interest in
calendars, specifically in querying calendar information. Aagin, to
store RDF efficiently datatyping support would be very useful.

Finally, querying in itself could benefit from knowledge about which
properies are likely to be present in a dataset and in what
proportions. a query will fail if you are not sure of the shape of the
data available, and minCardinality could help with that. 

So I guess to answer the questions Jim posed,

> * How would you test to see if the tools could support your needs?

I guess I would try to use them to speed up my RDF database and queries,
using some interesting sample adta.

> 
> * What would you want to do with this language?
>

Like Dan C and Mike, our efforts have focussed on describing people,
organisations and events, and the connections between them.
 
> * What features of the language would you look for and why (again, 
> please make your arguments in terms of using the language, not in 
> terms of the "aesthetics" of the design)?
> 

see above.

cheers

libby

Received on Thursday, 29 November 2001 09:26:25 UTC