Change Propagation in an Axiomatic Model of Schema Evolution for Objectbase Management Systems

Schema evolution is an important component of advanced information systems such as objectbase management systems. These systems typically support volatile and complex application domains that include engineering design, CAD/CAM, multimedia, and geo-information systems. The schema of these applications must be able to evolve along with the changing environment. There are two problems to consider in schema evolution: (i) semantics of change and (ii) change propagation. The first deals with the effects of the schema change on the overall type system. For example, the deletion of a property in a type affects the subtypes inheriting that property. Our previous work has introduced a sound and complete axiomatic model to deal with the semantics of change problem. The second problem deals with the techniques for propagating schema changes to the underlying objects. For example, the addition of an attribute to a type requires additional memory to be allocated to the objects so that values for the attribute may be stored. The first step of change propagation is to identify the affected objects. Subsequent steps carry out the actual changes. This paper deals with the first step by extending the axiomatic model with semantics to determine a sound and complete set of objects affected by a schema change. The extended model can be used with any method for carrying out the changes such as the conversion, screening, and filtering approaches proposed in the literature.