I agree with this statement, but I would add that it is very
important to distinguish between the core data models of XML
(infoset, PSVI, XDM) and other data models it is able to support
indirectly.
XML primarily -- natively -- has a tree-based data model.
It can store graphs, but doing so requires an extra data model such
as RDF or XBRL linkbases (which massively use XLink), and such data
models are not specifically bound or restricted to XML syntax.
ID/IDREF can be used for graphs... a native facility. You don't really
need RDF, etc.
