4.6 EquivalentTo issue

The EquivalentTo issue raises questions about what happens when things
of different types are said to be equivalent to each other (e.g., a
class is said to be equivalent to a property or an instance). I think
first we must determine whether class, property and instance are really
disjoint. For example, one of our requirements is "Classes as
instances." To satisfy this requirement, classes cannot be disjoint from
instances. Furthermore, in RDF Schema, everything is an instance. So
choosing to make these three concepts disjoint would be modifying the
semantics of RDF, and I get the feeling that some members of the group
would rather not do this. Still, the semantics are easier if we can make
class, properties, and instances disjoint. I am not really taking a side
on this issue now, but merely want to point out that 4.6 cannot be
resolved until it is.

If we decide that classes, properties, and individuals are not disjoint,
then there is no issue. Saying that instance a is of one type, b is of
another type and a equivalentTo b simply means that this object is of
both types.

What if we do decide that 2 or more of these types are disjoint? Well in
that case then you can have an inconsistent theory. It should be no
different than if you said that any two instances of disjoint classes
were equivalent.

Jeff

Received on Wednesday, 29 May 2002 10:00:19 UTC