A pattern-matching approach for automated scenario-driven testing of structured computational policy

Abstract

Organizations are policy-driven entities. Policy bases can be very large and complex; these factors are in the dynamic nature of policy evolution. The mechanical aspects of policy modification and assurance of the consistency, completeness, and correctness of a policy base can be automated to some degree. Such support is known as computer support for policy. We developed an object-oriented schema-based approach to structure policy. Our structural model consists of Unified Modeling Language class and collaboration diagrams. The structural model is used by a suite of testing tools. We present a case study to illustrate our approach to automated testing of policy. Our approach to test-case generation is based on the use of pattersn within policy statements and relationships between policy objects. The test spectrum has query-specific tests at one end, and the generic types of tests at the other end. We introduce the use of statistical inference to reuse test cases by determining the patterns that approximate the query-to-be-executed. Query mapping, anytime reasoning and fuzzy logic concepts in policies and their applications are discussed.