The big lie about the semantic web is this ridiculous notion that it is machine readable just because it has an unambiguous grammar to parser and has ontologies to disambiguate homonyms.
Thanks, I was rather involved in the research community. It has little to do with parsing text NLP-style. It has everything to do with annotating documents with formally structured KR data, or embedding the same into them.
In fact, my own involvement in the field started with Jim Hendler suggesting I look into the possibility of developing a spider that would wander across HTML pages and glean knowledge from them with an NLP technique. I worked on that for a while, then abandoned it and proposed to him instead that web pages should be marked up with formally parseable structured semantic data. Why parse a student's page saying he "Goes to U Maryland" when he could just write <claim obj1="me" obj2="UMaryland" rel="attend" ontology="http://ontology.org/university-ontology/">. That was 1995.