Transformation of deductive and inductive tasks between models of logic programming with imperfect information

Abstrakt

We analyze different models of imperfect information. We show a transformation under which fuzzy logic programs and generalized annotated programs (GAP) have the same computed and correct answers and Herbrand models.

We show that induction task for GAP can be transformed to multiple use of classical ILP with additional nomonoticity assumption in the background konwledge. We analyze conditions under which GAP can be transformed to Bayesian networks and vice versa.