| View and schema management for large software systems is a difficult task, further complicated by the use of multiple languages and data representations. An efficient representation supporting schema integration and testing is desirable.;This thesis develops a schema representation as a network of graphs, based on a conceptual data model (CDM) which includes concepts common to most current object-oriented and semantic data models and languages. CDM is class-based, and includes notions of typing restrictions, functional dependencies, and equations, all based on property paths. A graph network representing a schema is developed using algorithms that guarantee the soundness of the constraints modeled.;A complete axiomatization of the constraints is developed for the case in which the generalization hierarchy forms a lattice, and an example is given illustrating the difficulty in developing a complete axiomatization without the restriction. Decision procedures that involve straightforward navigation of the network representation are presented for various constraint membership problems. The procedures assume the problem schema is consistent, that it satisfies a stratification condition over the equational constraints, and that each functional dependency included in the schema is a form of complex or embedded key.;Applications of the theory are illustrated, including schema capture and design, view and query representation, query optimization, and join order selection. |
