Ontologies for Resolution Policy Definition and Policy Conflict Detection

Abstract

In previous work, the author devised a collection of ontologies to model the generic structure and characteristics of the APPEL policy description language [18] utilised within the ACCENT policy system [1]. Two ontologies, namely genpol.owl and wizpol.owl, were defined using OWL (Web Ontology Language [10]) to describe the generic aspects of the policy language and aspects of how the policy wizard (user interface) uses the language. These generic ontologies are explained in CSM-169 [3], while an ontology modelling a domain-specific implementation of the policy language for call control is described in CSM-170 [4]. This document describes how these ontologies have been extended to define the structure of resolution policies, in addition to standard domain policies. A resolution policy has a similar structural composition to a standard policy, but places restrictions on the characteristics of its components. While a standard policy is used to define how events within the domain are handled, a resolution policy is defined purposely to resolve run-time conflicts between standard domain policies. Conflict arises between a pair of standard domain policies whose actions clash if executed simultaneously. A resolution policy specifies the action to be taken when such conflict occurs. This report distinguishes between a standard policy and a resolution policy in APPEL, outlining the ontology description for each and highlighting subtle differences in their form. In particular, it demonstrates extensions to genpol and wizpol to specify generic resolution policies. Based on the generic extensions, callcontrol.owl was also expanded to include generic call control resolution actions. The call control ontology is described here as a concrete example of a domain-specific resolution policy language definition for the purposes of managing call conflicts. In addition, the report describes generic and domain-specific ontology extensions to aid in policy conflict detection using a filtering technique.