I thoroughly enjoyed the logic programming class, which was logic theory coupled with fun Prolog puzzles. I think Prolog is extremely well suited for pure algorithmic problems of a complex domain, as it allows encoding domain information and constraints efficiently.
I think so too. I use Prolog for data analysis now because it allows me to include both symbolic knowledge and numeric information. If the numeric modelling elements are straightforward, I code them in Prolog, and if they are not I call out to other things. The state of Prolog is such that I have to deal with a lot of integration/API/library related friction as a result, but I think that the benefits are worth it.