Re: RDFS bug "A property can have at most one range property"

> > If we talk only about RDF and RDF Schema, why
> > is better the new approach than the old one?
> Because the Semantic Web is a very different environment to a traditional
> knowledge- or data-representation system.  It is not possible to control, or
> even to discover, who will use, amend or extend your vocabulary.  Therefore,
> it is important to create specifications that are well-behaved in such
> circumstances.  In the example I gave previously, I argued that an
> interpretation as conjunction was 'better-behaved' than an interpretation as
> disjunction, because it would be more obvious sooner when there were
> problems with combinations of amended and/or extended vocabularies.  As a
> developer (of software, of systems or of vocabularies), I'd much rather see
> an easy-to-find, large problem that can be solved than a dificult-to-find
> intermittent problem that is never really traced successfully.
>                 - Peter

As a developer you have to solve problems, and application problems will
more often than not require to deal with integrity constraints (see my
earlier email for the relation to the conjunctive/disjunctive question
if you like). Now the developers tasks would be to make the best out of
it (check integrity based on whatever knowledge you have, look for
additional knowledge where possible or required (for example by loading
external schemata) and build awareness of the limitations of your
deductions due to potential incompleteness of knowledge into your
system). I understand that you'd rather like to have an easy-to-find,
solvable problem. However, remember that developing application rarely
requires to follow the wish-list of the deloper, more often it is the
wish-list of the guy who pays you (ideally driven by well-motivated
systems requirements) that sets the stage.

	Xonar GmbH

Received on Monday, 19 November 2001 06:59:06 UTC