OWL-POLAR: A Framework for Semantic Policy Representation and Reasoning

Abstract

In a distributed system, the actions of one component may lead to severe failures in the system as a whole. To govern such systems, constraints are placed on the behaviour of components to avoid such undesirable actions. Policies or norms are declarations of soft constraints regulating what is prohibited, permitted or obliged within a distributed sys- tem. These constraints provide systems-level means to mitigate against failures. A few machine-processable representations for policies have been proposed, but they tend to be either limited in the types of policies that can be expressed or are limited by the complex- ity of associated reasoning mechanisms. In this paper, we present a language that suffi- ciently expresses the types of policies essential in practical systems, and which enables both policy-governed decision-making and policy analysis within the bounds of decidabil- ity. We then propose an OWL-based representation of policies that meets these criteria and reasoning mechanisms that use a novel combination of ontology consistency checking and query answering. The proposed policy representation and reasoning mechanisms al- low development of distributed agent-based systems that operate flexibly and effectively in policy-constrainted environments.