As you might know already Tracker is now using RDF and SPARQL and also the Nepomuk ontologies. One of their most active developers is Phillip van Hoof. He just wrote two blogs on this topic: Introduction to RDF and SPARQL and More introduction to RDF and SPARQL. He gives a very nice overview of semantic data on the desktop. Of course, his examples are a bit Tracker specific but most can also be performed using sopranocmd.