Related pages:
The majority of examples in this book is shown in the Turtle serialisation of RDF,
for readability;
however, if data is published using just one serialisation format,
this should be RDF/XML,
as it is widely supported by tools that consume Linked Data.
The simplicity of N-Triples makes it popular for teaching people about RDF, and some parsers can read it more quickly because they have less work to do,
but the format's verbosity makes it less popular than the other formats.
.. once Turtle becomes a W3C standard, we'll see less and less use of RDF/XML. AS of this writing,
though, it's still the only standardized RDF serialization format.
If your're defining a fairly simple vocabulary primarily for your own use... and if your're concerned primarily with the striped nature of RDF/XML, you'll most likely want to just define your vocabulary in RDF and RDFS.
However, if you're documenting a model of a specific domain and you hope to encourage others to use it and, best of all, be able to use your data to make sophisticated queries, you're goint to want to use OWL to make advantage of its many inferential enhancements.