Well-Behaved IDL Theories

The field of nonmonotonic logic, sixteen years old now, is devoted to solve the problem of reasoning under incomplete knowledge, whose good understanding is essential to the construction of AI as a science and whose relevance reaches far beyond AI applications. During these years, many insights have been accumulated in the form of desirable properties the proposed formalisms should exhibit and of criticisms on the available solutions. This paper takes advantage on this experience to derive from them a sort of canon to be imposed to nonmonotonic formalisms. This canon is translated as a set of etiquette rules guiding knowledge representation into theories framed within the Inconsistent Default Logic, IDL. It is then established the important result that IDL produces a unique extension for a theory constructed according to these rules. This result calls forth IDL as an interesting alternative to credulous common sense reasoning formalization fulfilling many desired properties.