Database Reference
In-Depth Information
We believe that the use of the framework in Figure 2 and any extensions of it (par-
ticularly when used together with Wand and Weber's (2002) broader framework) would
provide future human factors research on conceptual data modeling a stronger foundation
and give the researchers an incentive to specify the relationships between the variables
of interest at a more detailed level and present them better in relation to other, potentially
signifi cant variables.
New Areas of Focus
Finally, we would like to propose two additional foci for conceptual data modeling
research: a) basic research on concept formulation, categorization, and usage, and b) applied
research on data modeling processes.
First, as Wand and Weber (2002) point out, we need a better understanding of the
psychological processes in data modeling and the ways the tools affect these processes.
This will enable us to fi nd a fi rm theoretical basis for human factors research on data mod-
eling. Researchers in this area should be interested not only in the characteristics of the
current models, but the reasons underlying the potential performance differences between
various approaches to data modeling. Batra's (1993) framework of error behaviors and the
introduction of the GEMS model to this domain by Batra and Antony (2001) are excellent
steps in the right direction. As Siau (1999) points out, cognitive science is potentially a very
useful reference discipline, especially the research in cognitive science that has its roots
in cognitive psychology or in artifi cial intelligence (Batra, 1993; Henderson & Peterson,
1992; Rosch & Mervis, 1975; Rosch, Mervis, Gray, Johnson & Boyes-Braem, 1976; Smith
& Medlin, 1981). Applied research in this fi eld has been done, for example, in marketing
and organizational behavior (for representative examples see Day & Lord, 1992; Fiol &
Huff, 1992; Ozanne, Brucks & Grewal, 1992).
It is essential to point out here the very signifi cant general theoretical foundation work
Wand and Weber have done on several dimensions of conceptual modeling, particularly in
demonstrating how ontology can be used as a basis for conceptual analysis. This research
has been published in journal articles (Wand et al., 1995; Wand, Storey & Weber, 1999;
Wand & Weber, 1993, 1995, 2002) and as a monograph (Weber, 1997). We believe that
their work is an invaluable foundation for future conceptual and empirical work in this area,
including the work on usability.
The essence of all modeling is in the identifi cation of concepts and categorization of
them (Booch, 1994, Chapters 1-4; Coad & Yourdon, 1991, Chapter 1). The links between
theoretical research on categorization and data modeling are still somewhat weakly defi ned,
although Parsons and Wand's (Parsons, 2003; Parsons & Wand, 1997, 2000) work is a very
important contribution and an excellent example of the type of research that is needed in
this area. An additional important contribution would be a conceptual analysis of the char-
acteristics of various data modeling techniques compared with categorization theories (see
Henderson and Peterson (1992) for a concise introduction) and an empirical verifi cation of
the results of this research. The central focus of this research should be on the relationships
between individual abilities, individuals' histories, situation characteristics, perceptions of
reality, and categorization behavior. On the other hand, it is very important to note that data
models are not (or at least should not be) created in a social vacuum; a data model describes
a collective cognitive view about an organization. If reality is socially constructed (Berger
& Luckmann, 1967) and information processing is greatly affected by social structures and
