The emergence of new data models requires further research into federated database systems. A federated database system (FDBS) provides uniform access to multiple heterogeneous databases. Most FDBS's provide access to only the older data models such as relational, hierarchical, and network models.

A federated system requires a meta data model. The meta model is a uniform data model through which users access data regardless of the data model of the data's native database.

This dissertation examines the question of meta models for use in an FDBS that provides access to relational, object oriented, and rule based databases. This dissertation proposes Cyrano, a hybrid of object oriented and rule based data models. The dissertation demonstrates that Cyrano is suitable as a meta model by showing that Cyrano satisfies the following three criteria:

1. Cyrano fully supports relational, object oriented, and rule based member data models.

2. Cyrano provides sufficient capabilities to support integration of heterogeneous databases.

3. Cyrano can be implemented as the meta model of an operational FDBS.

This dissertation describes four primary products of this research:

1. The dissertation presents Cyrano, a meta model designed as part of this research that supports both the older and the newer data models. Cyrano is an example of analytic object orientation. Analytic object orientation is a conceptual approach that combines elements of object oriented and rule based data models.

2. The dissertation describes Roxanne, a proof-of-concept FDBS that uses Cyrano as its meta model.

3. The dissertation proposes a set of criteria for the evaluation of meta models. The dissertation uses these criteria to demonstrate Cyrano's Suitability as a meta model.

4. The dissertation presents an object oriented FDBS reference architecture suitable for use in describing and designing an FDBS.