A Design Theory for Solving the Anomalies Problem
Edward P. F. Chan · SIAM Journal on Computing · 1989
A theory is proposed for designing database schemes that are free of update anomalies. Unlike previous approaches, insertion and deletion anomalies are investigated in the context of a relation scheme, while replacement anomalies are studied in the context of a database scheme. Two simple models are developed for analyzing when a relation scheme is free of insertion and deletion anomalies. Techniques are also proposed for obtaining desirable decompositions that are free of insertion and deletion anomalies. A class of database schemes that is free of replacement anomalies is also proposed. This class of schemes is highly desirable with respect to constraint enforcement when attribute values of some tuple are being changed. By making different assumptions on the modifiable attributes, several important classes of database schemes that are free of replacement anomalies are characterized. Throughout, we assume update operations are performed on relation schemes at the conceptual level.