RE: Explaining why we use RDF instead of just XML

> I think everything that can be represented with xml can be represented
> with rdf but not the other way round. Or is there a way in plain xml to
> express a circular relation like "john dislikes peter and peter
> dislikes john"?

Yes. For one thing the set of valid RDF/XML documents is of course a subset
of the set of valid XML documents, so the fact that it can be done in
RDF/XML proves that something can be done in XML. Whether it can be done in
"plain" XML depends on what you mean by "plain".

There are some things that can be expressed in RDF that can't be expressed
in RDF/XML (or are those issues resolved since by last proper reading of the
specs). However an XML format that found a way of expressing them wouldn't
be impossible.

The real difference is that RDF has a greater degree of meaning outside of
the application a given document was created by/for.

To compare deliberately inscrutable examples the XML snippet:

<aaa><bbb>yyy</bbb><ddd><eee fff="xxx"/></ddd></aaa> means absolutely
nothing (and to a machine there is precious little difference between
"<bbb>" and "<price>" or "<name>").

The same treated as RDF/XML means little also. However we have at least an
idea of how the structure is meant to be interpreted.
RDFS, DAML+OIL or OWL could give us the ability to learn that these unknown
elements are the same as, or related to, elements we do know, and hence
entail statements that are meaningful to us.

They also give us a good way to expand on the above and know how we can fit
our new wonderful <rrr> element that we've come up with where applications
would expect an <eee> (again to the machine there is precious little
difference in "<eee/><rrr/>" and "<user/><advancedUser/>").

Received on Thursday, 26 June 2003 07:04:09 UTC