On abstract modular inference systems and solvers

Integrating diverse formalisms into modular knowledge representation systems offers increased expressivity, modeling convenience, and computational benefits. We introduce the concepts of abstract inference modules and abstract modular inference systems to study general principles behind the design and analysis of model generating programs, or solvers, for integrated multi-logic systems. We show how modules and modular systems give rise to transition graphs, which are a natural and convenient representation of solvers, an idea pioneered by the SAT community. These graphs lend themselves well to extensions that capture such important solver design features as learning. In the paper, we consider two flavors of learning for modular formalisms, local and global. We illustrate our approach by showing how it applies to answer set programming, propositional logic, multi-logic systems based on these two formalisms and, more generally, to satisfiability modulo theories.