Viable Non-monotonic Applications

Non-monotonicity enhances the expressivity of a representation framework allowing high-level representations of (inherently non-monotonic) problems close to their natural specification. This gain in expressivity should not though compromise (severely) the computational effectiveness of the solutions based on non-monotonic representations. In this paper we argue that for abduction and consequently also for other non-monotonic frameworks to become viable for "real-life" applications it is important for these to be integrated with appropriate specialized constraint solvers for (low-level) domain constraints. It is thus important to define and implement efficiently appropriate interfaces between non-monotonic reasoning systems and suitable constraint solvers. We describe briefly an example of such an interface and discuss how it helps to develop viable applications. Such successful integrations will have the advantages of modularity and flexibility that the high-level non-monotonic representation offers together with computationally effective solutions to complex problems.