"Bayesian Belief Network" is just another term for a "Bayesian network". As far as I know, the big difference is when it comes time to make an inference. To inference on a Bayesian network requires all probabilities to be specified, whereas a Bayesian Knowledge Base (BKB) can handle incomplete information/uncertainty. One link to look at:

http://www.cs.indiana.edu/event/maics96/Proceedings/Borghetti/node1.html#SECTION00010000000000000000

Also search for Eugene Santos, Jr. (currently at Dartmouth) and Kathryn Laskey (GMU) -- they have written a lot about BKBs.