Entity-Relationship Meta-model Databases with Unknown Meta-Attribute Values

Abstract

In the entity-relationship (ER) meta-model, modeling constructs such as entities, relationships, attributes, etc., in the concrete conceptual model are all represented in meta-entity classes, and items of information which qualify the modeling constructs (e.g. whether an entity class in the concrete conceptual model is strong or weak) appear as meta-attributes, associated with a particular meta-entity class. In an agile software development process, conceptual model construction and the consequent database design often proceeds on the basis of incomplete requirements specification, especially in the early iterations, and the information on a model construct, i.e. the concrete value of a meta-attribute, is hard to predict and specify. In this thesis, we deal with the unknown values by extending the ER meta-model to allow meta-attributes to have the value unknown . These unknowns are taken into account in the physical database design. We have implemented the triggers to refactor the physical database when the unknown attribute values in the conceptual model become known. The thesis demonstrates that the database design based on the concrete conceptual model can be re-factored when missing information is identified and updated.