Fault-tolerant and approximate reasoning in multi-source environments

When different knowledge-based systems must cooperate to perform decision tasks that are beyond their individual capabilities, we are faced with the problem of combining knowledge in a multi-source environment. In particular, we are confronted with two main difficulties: the prospect of inconsistency, which arises when different knowledge bases are merged together, and the high computational complexity of reasoning with very large pools of combined information. In this paper, we define a formal framework which handles both aspects of consistency and tractability, and which is useful to specify knowledge retrievers. This framework tolerates inconsistency and enables a knowledge retriever to infer non-degenerative conclusions when conflicting viewpoints are combined. Furthermore, approximate reasoning is incorporated in order to perform efficient query answering using combined knowledge. Finally, a stepwise procedure is included for improving approximate answers and allowing their convergence to the right answer.