Using DEKAF to understand data modelling in the practitioner domain
- PDF / 184,198 Bytes
- 10 Pages / 594 x 797 pts Page_size
- 55 Downloads / 209 Views
1997 Operational Research Society Ltd. All rights reserved 0960-085X/97 $12.00
Using DEKAF to understand data modelling in the practitioner domain S Hitchman Faculty of Business and Social Studies, Cheltenham & Gloucester College of Higher Education, PO Box 220, The Park Campus, Cheltenham, GL50 2QF, UK Despite the ubiquitous use of entity-relationship modelling for more than twenty years, there is surprisingly little evidence of how effective data modelling is in the commercial domain, and this evidence suggests that modelling is problematic. This paper evaluates the literature on the effectiveness of data modelling in the practitioner domain, showing that implicit objectivist assumptions about narrative are questionable. A domain expert knowledge approach framework (DEKAF) is described in the context of overcoming problems of research generalisability. DEKAF provides both a useful way of understanding and thinking about the data modelling process and a way of making assumptions explicit in a particular practitioner domain. A summary of the findings of action research shows that DEKAF can be successfully used and can give insight into effective practitioner domain modelling.
Introduction In a discussion of a ‘mythical’ case study by Turner and Jenkins (1995, p 7), a project involving three European Union countries is described in which there are three different analysis methodologies and technical platforms “... one of the few areas of commonality between organizations is over the importance of data models in development.” Data modelling is ubiquitous in analysis methods, particularly entity-relationship modelling (ERM), although challenged by the so-called object-oriented models (OOM) and by object-role modelling (ORM, also known as FACT modelling and within the NIAM method). However, what little evidence there is about effective data modelling in the commercial domain (for example, Batra & Marakas, 1995; Hitchman, 1995a) points to data modelling being problematic. This paper argues that current research is limited in informing the debate about modelling and that, given the widespread use of data modelling, a better research framework needs to be established. Data modelling is studied from a theoretical viewpoint while also being carried out by practitioners and the literature can be characterised by distinguishing between texts aimed primarily at ‘theory’ (for example Batini et al, 1992) or at a ‘practitioner’ (for example Barker, 1989) audience; although particular texts are not necessarily exclusive. Veryard (1992), a text in a ‘practitioner series’, makes use of the work of ‘theorists’, but the reverse is not generally true. The practitioner literature tends to echo the literature concerning IT management consultancy, where “... much of the literature ... has tended to be of a descriptive and/or prescriptive nature
... fails to go beyond descriptions of what consultants themselves claim to do, and their prescriptions for best practice ...” (Bloomfield & Daieli, 1995, p 25). Lewis (1994) shows that there is
Data Loading...
