Narrow Results

We present an approach to knowledge compilation that transforms a function-free first-order Horn knowledge base to propositional logic. This form of compilation is important since the most efficient reasoning methods are defined for propositional logic, while knowledge is most conveniently expressed within a first-order language. To obtain compact propositional representations, we employ techniques from (ir)relevance reasoning as well as theory transformation via unfold/fold transformations. Application areas include diagnosis, planning, and vision. Preliminary experiments with a hypothetical reasoner indicate that our method may yield significant speed-ups.

Constraint Satisfaction Problems (CSPs) offer a powerful framework for representing a great variety of problems. The difficulty is that most of the requests associated with CSPs are NP-hard. When these requests have to be addressed online, Multivalued Decision Diagrams (MDDs) have been proposed as a way to compile CSPs.

In the present paper, we draw a compilation map of MDDs, in the spirit of the NNF compilation map, analyzing MDDs according to their succinctness and to their tractable transformations and queries. Deterministic ordered MDDs are a generalization of ordered binary decision diagrams to non-Boolean domains: unsurprisingly, they have similar capabilities. More interestingly, our study puts forward the interest of non-deterministic ordered MDDs: when restricted to Boolean domains, they capture OBDDs and DNFs as proper subsets and have performances close to those of DNNFs. The comparison to classical, deterministic MDDs shows that relaxing the determinism requirement leads to an increase in succinctness and allows more transformations to be satisfied in polynomial time (typically, the disjunctive ones). Experiments on random problems confirm the gain in succinctness.

Tractability versus expressiveness is a problem underlying the current generation of problem solvers. It is desirable to specify the problem, along with knowledge required for its solution, in the most expressive means possible. However, from the perspective of the solver, the most efficient representation is the least expressive one. One solution to this problem is to devise a means of transforming an expressive problem description into an efficient problem solver. Knowledge compilation is the process of transforming a domain theory into a form specialized for the solution of a given problem set. This paper will present OPCOMP, a system devised to compile a first order quantifier free predicate calculus domain theory into a problem reduction solver.