Combining hypothetical and nonmonotonic reasoning modes with application to updating-a possibilistic logic approach

In this paper we only consider possibilistic logic formulas which are weighted by lower bounds of a necessity measure. Thus, formulas in a knowledge base are rank-ordered according to their levels of certainty. These levels of certainty can be made context-dependent. Possibilistic logic has been shown to provide an efficient tool for handling uncertain and nonmonotonic reasoning, including hypothetical reasoning (when weights depend on context). The basic mechanisms of possibilistic logic are first recalled, emphasizing its ability to handling partial inconsistency. The combination of these reasoning capabilities enables us to perform a case-by-case reasoning. This is applied to updating problems where default rules describe the way the world is evolving with time and where initial states may be incompletely known, thus requiring a case-by-case reasoning in order to compute the following states which are possible. Possibilistic logic is here shown to provide mechanisms not only for minimizing the abnormality (the higher the weight of a formula, the more abnormal are the situations which contradict this formula), but also for letting some formulas vary with context, as in circumscription-based methods