Armstrong databases: validation, communication and consolidation of conceptual models with perfect test data

Abstract

Conceptual models are relational database schemata that result from conceptual data modeling. Conceptual models usually capture the semantics of the underlying application domain inadequately. Therefore, the structure of the conceptual model is often inadequate, too. Academic and commercial database design tools often advocate the use of good test data to validate the adequacy of the conceptual models they produce. In this article we provide evidence that Armstrong databases constitute perfect test data. In particular, Armstrong databases capture perfectly the perceptions of the design team about the semantics of the application domain. Therefore, Armstrong databases serve as an excellent medium to validate and consolidate the understanding of an application domain’s semantics, and to communicate this understanding between different stakeholders of the target database. An overview is given about recent advancements on the structural and computational properties of Armstrong databases. These advancements suggest that Armstrong databases provide the foundations necessary to establish an agile database design methodology. Such a methodology complements existing approaches to database design, and is not meant to replace them.