Possibilistic conditional tables

Abstract

On the one hand possibility theory and possibilistic logic offer a powerful representation setting in artificial intelligence for handling uncertainty in a qualitative manner. On the other hand conditional tables (c-tables for short) and their probabilistic extension provide a well-known setting for representing respectively incomplete and uncertain information in relational databases. Although these two settings rely on the idea of possible worlds, they have been developed and used independently. This paper investigates the links between possibility theory, possibilistic logic and c-tables, before introducing possibilistic c-tables and discussing their relation with a recent certainty-based approach to uncertain databases and their differences with probabilistic c-tables.