I'd say that the most useful basic features of RDF are:Elements not defined in the Atom core must be namespace qualified (and in a different namespace than Atom, of course).
1) Property names are namespaced for extensibility.
An entry or feed can be referred to from arbitrary places, whether in or out of the feed where it lives, using the value found in its identity construct.2) Important entities can be assigned global identifiers so that they can be referred to externally.
Child elements say something about their parents. (This seems so intuitively obvious that I wouldn't think people would make XML that didn't work this way, but from the sound of it, there are examples out there of XML where child elements aren't related to their parents. I for one would understand the importance of making this explicit if I could see such examples).3) Statements are properties of an object rather than being simple name/value pairs.
4) The model has properties such as monotonicity: which results inNo element can change the meaning of its siblings.
statements having a fixed meaning that is not negated by the presence
of other statements; this makes it possible to mix vocabularies
without needing lots of coordination to prevent them from conflicting.