I think that generalization is false as often as it is true.
But also, have you read any of the papers involved? Datalog is pretty simple. It's a restricted, forward chaining prolog. Once you know that, you can recreate most of it from that description alone.
