DATA MODEL QUALITY

An initial starting point for ensuring that data is of good quality is to have a conceptual data model that is both complete and correct. There are a number of approaches that may be taken to the assessment of the quality of a conceptual data model; some of these approaches are purely qualitative, whilst others are quantitative, applying statistical techniques to the numbers of entities, attributes and relationships in the model. Many of these approaches have been reviewed in (Genero and Piattini, 2002).

An easily applied qualitative model for the assessment of the quality of a data model amongst those reviewed by Genero and Piattini is that proposed by (Reingruber and Gregory, 1994), which is shown in Figure 7.2.

FIGURE 7.2 The five dimensions of data model quality

In this model, Reingruber and Gregory have augmented the correctness and completeness dimensions with two further orthogonal dimensions – the syntactic dimension and the semantic dimension. The syntactic dimension addresses how the modelling language and its syntax have been used whilst the semantic dimension addresses the relationship between the model and the data requirements of the business area that the model represents. Applying these orthogonal dimensions together, we get the four dimensions of syntactic correctness, syntactic completeness, conceptual correctness and conceptual completeness. Reingruber and Gregory have added a fifth, overarching, dimension that they call enterprise awareness. This recognises that any data model for a specific business area or set of business processes should be seen as a subset of the enterprise or corporate data model. It is the enterprise awareness dimension that is most often overlooked by data modellers working as part of project teams involved in the development of information systems.