Answer Set Programming: Towards Efficient and Scalable Knowledge Representation and Reasoning
Papers from the AAAI Spring Symposium
Alessandro Provetti and Tran Cao Son,Cochairs
Answer set programming (ASP) is the realization of much theoretical work in nonmonotonic reasoning, AI, and logic programming over the last 12 years. It is based on the view of program statements as constraints on the solution of a given problem. Subsequently, each model of the program encodes a solution to the problem itself. For instance, an ASP program encoding a planning scenario has as many models as valid plans. This schema is similar to that underlying the application of SAT algorithms to AI and, in fact, the ranges of applicability of these two techniques are similar. However, thanks to the inherent causal aspect of answer set semantics, we can represent default assumptions, constraints, uncertainty and nondeterminism in a direct way. Several ASP systems are now available, (such as DeReS, dlv, smodels and XSB); they support provably correct inferences and are at least as fast and scalable as SAT checkers. These exciting results for the NMR community are attracting the attention of researchers in fields such as planning, cryptography and system verification. Participants addressed questions such as: "What are the strengths of ASP vis-a-vis satisfiability, CSP, abduction, argument- based reasoning and model checking?" "What applications are go-ing to give a perceivable edge (diagno-sis, workflow, configuration ...)?" "What is a fair benchmark to evaluate progress in implementations?"