clp(pdf(y)): Constraints for Probabilistic Reasoning in Logic Programming

* Final gross prices may vary according to local VAT.

Get Access


We argue that the clp(X) framework is a suitable vehicle for extending logic programming (LP) with probabilistic reasoning. This paper presents such a generic framework, clp(pdf(Y)), and proposes two promising instances. The first provides a seamless integration of Bayesian Networks, while the second defines distributions over variables and employs conditional constraints over predicates. The generic methodology is based on attaching probability distributions over finite domains. We illustrate computational benefits of this approach by comparing program performances with a clp(fd) program on a cryptographic problem.