Service oriented reaction rule agent

The thesis discusses the design and implementation of service-oriented reaction rule agents in enterprise business environments. Business rule systems allow business policies to be created at run time by business knowledge workers. Enterprise applications using rule based designs can be more adaptive to the fast changing business requirements. The service oriented architecture abstracts functionalities of enterprise applications as self-contained, self-describing services that are discoverable and consumable through standard protocols. The architecture shows great potentials in enterprise application integrations. The service oriented architecture may fundamentally change how enterprise applications are designed, deployed and integrated. In this thesis, we propose the concept of the service oriented reaction rule agent to capture the essence of the business rule oriented design paradigm and the service oriented architecture. We define the service oriented reaction rule agent's data flow framework using Agent Object Relationship Modeling (AORM). Based on the framework, we present the general reaction rule markup language model that can be easily extended into specific business domains. We further identity the key design patterns and architectural patterns in implementing the service oriented reaction rule agents and the usage of hierarchical cache design pattern in improving their general performance. We give the use case study of using the service oriented reaction rule agent in federated single sign-on. We also analyze the computational complexity of the reaction rule agent and its architecture properties.