5. Discussion 4. an Example 3.3 Uncertain Information 3.2 Inconsistent Information 3. a General Framework for Managing Imperfect Information 3.1 Incomplete Information

Many different models for handling imperfect information have been proposed in the context both of databases and of artificial intelligence. In general these are either symbolic methods for handling incomplete information or numerical models for handling imprecision, vagueness or uncertainty, and all typically handle a different aspect of the problem. In this paper we suggest that in order to properly model imperfect information in deductive database we should adopt an eclectic position that attempts, in a principled way, to use whatever model best fits the given data. To do this we suggest a general framework based upon previous work on a system of argumentation.