Measurement of the Quality of an FCO-IM Conceptual Data Model

Abstract

Data models provide the foundation to organization's activities since they support the organization's systems and data. Therefore, the quality of the data models is foremost. We describe a methodology to measure the quality of conceptual data models created using a fact oriented data modeling called Fully Communication Oriented Information Modeling (FCO-IM). The measurement method is based on the framework to measure the quality of conceptual model by Lindland et al. Four components are to be considered in the measurement: domain, model, language, and audience interpretation. The quality are measured on three aspects: syntactic quality (measured by syntax correctness), semantic quality (measured by feasible validity and feasible completeness), and pragmatic quality (measured by feasible comprehension). The method is then used to determine the quality of several FCO-IM conceptual data models that were created using a pattern language of conceptual data models, a new method in data modeling that we are currently researching. The method contributes in data modeling area by providing a quantitative and instructive way of measuring the quality of conceptual data models, especially in FCO-IM. Keywords: conceptual model, conceptual data model, data modeling, FCO-IM, measurement, pattern, quality IntroductionInformation is one of the critical assets of a modern organization. Information is extracted from data stored in database systems. An aspect of data management is the definition of the structures of data. The structures of data are designed in an activity called data modeling and the results are data models. Data models provide the foundation to organization's activities since they support the organization's systems and data [20]. Therefore, the quality of the data models is foremost.Data model is a collection of conceptual tools for describing data, data relationships, data semantics, and consistency constraints [16]. Three levels of data models are defined [19]: conceptual, logical, and physical data model. Conceptual data model is a relatively technologyindependent specification of data structures and is close to business requirements [19]. We focus on conceptual data model rather than the logical or physical data model because a conceptual data model can be viewed as the translation of business requirements into technical form of the structures of data (thus, it serves a "link" between human and machine) and providing logical and physical data model is a matter of transforming the conceptual data model using established algorithms. Thus, the challenge is how to provide high quality conceptual data models.