[
Lists Home |
Date Index |
Thread Index
]
On Tue, 21 Dec 2004 14:23:52 -0500, Elliotte Harold
<elharo@metalab.unc.edu> wrote:
> Roger L. Costello wrote:
>
> > Does this indicate that Approach #2 is superior? Is entropy a way to
> > measure the value (quality) of Schemas? /Roger
>
> The problem with measuring entropy by schema is like that of measuring
> entropy by the shortest program that can produce a given string. If
> someone devises a shorter program or a more explicit schema, then the
> entropy decreases. In this case, it's straightforward to construct a
> more explicit schema that restricts the RDF document as much as the
> other schema restricts the XML document:
<snip>example Schema</snip>
> I would claim that if it's possible to write XSLT stylesheets that
> losslessly convert between two formats in both directions, then the two
> formats are isomorphic and the entropy does not change from an instance
> document in one format to the instance document in the other format.
I think your explanation makes it clear that you can't measure the
information entropy of a just a component of the system, it only makes
sense to measure the information entropy of the entire system: we
might be able to say that a given Schema constrains the information
contained by a given instance document. However, that same instance
information is only constrained if the Schema is itself available and
enforced. That tends to suggest that the answer to Roger's original
question is no; unless you know how the rest of the system works in
conjunction with both the Schema and the instance document you still
don't know what the total information entropy of the system is?
--
Peter Hunsberger
|
