A systematic database summary generation using the distributed query discovery system

Abstract

This paper introduces an approach to generate a database summary systematically using the distributed query discovery system, MASSON. Our approach is first to create an object-view and partition the database based on the object-view into clusters with similar properties, and then to generate the summary for each cluster. For this purpose, we propose a data set representation framework and introduce a proper similarity measure framework. The paper also describes the techniques used to generalize the generated primitive summary descriptions by MASSON and to improve the performance of the system using clustered computers and CORBA. © Springer-Verlag Berlin Heidelberg 2004.