[
Lists Home |
Date Index |
Thread Index
]
Jonathan Robie wrote:
>
> The Formal Semantics is all about structural subsumption, not about
> validation.
In general there are two things about type or classification systems:
1) a way to determine whether individuals (or objects) have a particular
type. In the so-called "static type" usages, an individual (or object) is
explicitly declared to be a member of a particular class i.e. have a given
type. In dynamic or run type usages, the system inspects the individual and
classifies it. This is akin to validation, but as you say is not all that
classifications are.
2) classes may be declared to be subclasses of other classes. Class
hierarchies may be built using logical combinators which generally implement
what XML Schema calls "derivation by restriction" and "derivation by
extension". The logical combinators include "intersection" by which a
subclass is defined to be the intersection of a set of super classes
(derivation by restriction), "union" where a subclass is defined to be the
union of a set of super classes (derivation by extention) etc.
> In general, it tries to define structural subsumption so that
> when it looks at a query expression and a schema, the conclusions it draws
> would be compatible with the conclusions a schema processor would make
when
> validating instances from the universe of documents the schema defines.
Having such a class hierarchy as described above does indeed allow one to
draw logical inferences about the classes and class membership. Such
inferences are indeed useful. Often one might not need to attempt to
classify or validate a set of individuals (or objects) against a class
definition, when such inferences can be drawn -- this is akin to query
optimization.
Jonathan
|
