Distributed Defeasible Reasoning in Ambient Intelligence
Antonis Bikakis, Grigoris Antoniou, Panayiotis Hassapis · 2008
Ambient Computing environments host various agents that collect, process, change and share the available context information. The imperfect nature of context, the open and dynamic nature of ambient environments, the different viewpoints from which the ambient agents face the same context, and their heterogeneity with respect to the language and inference system that they use, have introduced new challenges in the study of Distributed AI. The current paper presents a knowledge representation model based on the Multi-Context Systems paradigm that handles these requirements by modeling ambient agents as peers in a P2P system, local context knowledge as peer rule theories, and mapping rules, through which the ambient agents exchange context information, as defeasible rules. To resolve potential inconsistencies that may arise from the interaction of local theories through the mappings (global conflicts), the proposed method uses a preference relation on the system peers, which may express the trust that an agent has in the knowledge imported by other agents. On top of this model, we have developed four alternative strategies for global conflicts resolution, which differ in the type and extent of context knowledge that the ambient agents exchange in order to evaluate the quality of the imported context information. The four strategies have been respectively implemented in four versions of a distributed reasoning algorithm for query evaluation in Multi-Context Systems.