Abstract
The summarization system takes a Database (DB) table as input and produces a reduced version of this table through both are writing and a generalization process. The resulting table provides records with less precision than the original but it is very informative of the actual DB content. This reduced form can be used as input for advanced Data Mining processes. Several approaches of DB summarization have been proposed in the literature. The most recent is the SaintEtiQ summarization model, proposed initially by Raschia.(1) Based on a hierarchical conceptual clustering algorithm, SaintEtiQ builds a summary hierarchy from DB records. In this paper, we propose to extend this DB summarization model by introducing some optimization processes including: (i) minimization of the expert risks domain, (iii) building of the summary hierarchy from DB records, and (iv) cooperation with the user by giving him summaries in different hierarchy levels.