Royer introduces a formalism to handle incomplete information in deductive databases. The formalism is based on preference relations among the models of the underlying database. Two examples of preferential semantics illustrate the approach: the perfect model semantics and the making choices semantics.

This research work’s main contribution is that it provides a flexible database design method, due to the sharp separation between the database specification and the control of incomplete information. Of course, much work remains to be done for query evaluation in knowledge bases and databases. The method outlined by this paper seems promising.