Schema Integration Meta-Knowledge Classification and Reuse

Recent progress in communication and database technologies has drastically changed user data processing capabilities. The present situation is characterised by a growing number of applications that require the ability to access and manipulate data from various pre-existing database sources. A possible solution to deal with multiple databases is to logically integrate the component databases by the provision of one or more tailored global schemas. The end-user therefore is presented with a homogeneous and consistent view of the (possibly heterogeneous) multidatabase. Schema integration is an involved process that requires the supervised detection of similarities and dissimilarities between the component schemas, reconciling the dissimilarities once detected and providing a homogeneous view to the end-user. We are currently investigating the schema integration problem in object-oriented multidatabase systems. This paper describes one aspect of our work, namely, the classification and reuse of the knowledge accruing from the schema integration process. Such knowledge is classified into the knowledge that accrues from detecting and reconciling semantic heterogeneity between local schemas and the knowledge that accrues from generating the global schemas. We argue that reuse in schema integration is a vital issue and proves most valuable in saving the integrator's effort, simplifying the generation of new global schemas and supporting global schema evolution.