Reasoning Under Incomplete Information In Artificial Intelligence: A Comparison of Formalisms Using a Single Example (Wiley Professional Computing)

The formalization of ``revisable reasoning'' has been the object of numerous works, developed independently and using many diverse approaches--approaches that are purely symbolic, use numbers to quantify uncertainty, are close to formal logic or less formalized; some deal with exceptions, and a smaller number consider the problem of knowledge bases of revision. This work presents and compares several of these revisable (incomplete) reasoning methods for use in AI. Each method is systematically evaluated with a single example to give the reader an appreciation of the rationale and use of each formulation. The logics considered include: default logic, non-monotonic modal logics, the supposition-based logic, the conditional logics, and the logics of uncertainty. The book also discusses the contribution of works on truth maintenance and logic of action.