A Local Search Approach to Modelling and Solving Interval Algebra Problems

摘要

Local search techniques have attracted considerable interest in the artificial intelligence community since the development of GSAT and the min-conflicts heuristic for solving propositional satisfiability (SAT) problems and binary constraint satisfaction problems (CSPs) respectively. Newer techniques, such as the discrete Langrangian method (DLM), have significantly improved on GSAT and can also be applied to general constraint satisfaction and optimization. However, local search has yet to be successfully employed in solving temporal constraint satisfaction problems (TCSPs). This paper argues that current formalisms for representing TCSPs are inappropriate for a local search approach, and proposes an alternative CSP-based end-point ordering model for temporal reasoning. The paper looks at modelling and solving problems formulated using Allen's interval algebra (IA) and proposes a new constraint weighting algorithm derived from DLM. Using a set of randomly generated IA problems, it is shown that local search outperforms existing consistency-enforcing algorithms on those problems that the existing techniques find most difficult.