tom,
those that know what they're talking about when they say "social graph" (admittedly, not everyone that uses the term) are talking about the application of general mathematical "graph theory" (see wikipedia) to the interconnections of bi-directional "friend"-type link networks.
mathematicians have been using the term "graph" for decades, if not centuries, with this meaning. to completely dumb it down, a "semantic graph" would be what a computer (ie. Google) uses as in internal representation of the "semantic links" (uni- or bi-directional, direct or indirect, etc.) to do its magic.
i can understand how a LibArts major like yourself wouldn't have come into contact with these terms in your schooling, but please take this opportunity to check out the "dark side" of the maths behind this stuff if you're at all interested. or maybe you only like thinking about the end-user experience and not the underlying processes involved in making it work?? no harm, no foul. some people like both though, and the term "graph" is accurate for what they're working on, when described to anyone.
now "web 3.0"... that's just idiocy.

Not only does it falsely imply the need for a "social web" (web 2.0) to predate or underly the "semantic web", it also assumes that there will be no other major improvements prior to the realization of a "semantic web".
*sigh*