On applying classification to schema integration

Abstract

Discusses how the concepts of subsumption and classification (as described in the KL-ONE family of systems) can be adapted to automatically identify various relationships among classes. Classes can overlap, or be equivalent or disjoint, or one class may include another class. Computing such relationships is based on the 'subsume' function, which determines whether a given class is a superclass of another class. Relationships thus identified can be used to merge classes of two or more schemas (i.e., integrating these schemas) with the help of the classification function and other schema restructuring operators.<>