3.5    A Taxonomy of Enterprise Modeling Based on Purpose

Enterprise modeling may have many purposes, as indicated by the definitions in section and observations from the three initial empirical studies. By purpose of enterprise modeling is meant answering the question "What is the primary intended use of enterprise modeling?". Differences in purpose lead to differences in both approach to modeling and in qualities of the models that are developed. The main reason for introducing a taxonomy of enterprise modeling based on purpose is to avoid comparison and critique of approaches that have different purposes.

3.5.1 Three categories of enterprise modeling

Three main categories of purposes for enterprise modeling are proposed (adapted from Christensen et al., 1995:1175): The three categories are not intended to be complete and disjoint in the sense that any purpose of enterprise modeling fits one and only one of the categories. They are more intended as pragmatic categories.

One immediate reaction to the above categories may be "Only three?! I know of a lot more purposes!" A fundamental insight is that purposes depend upon each other, and while some purposes are direct, others are indirect. E.g., the purpose of "change the way work is done" is an indirect purpose: Creating models does not change the way work is done. Creating models may change your understanding of some situation, leading to change of how work is done. Hence, understanding precedes change.

Improved understanding can be a result of modeling belonging to all three categories. Category I is exclusively dedicated to sense-making and communication. Category II may increase understanding of phenomena that are not obvious (e.g., dynamic behavior of complex phenomena), and modeling in category III may guide a novice in his work and thereby improve his understanding of how work is done.

Category I: Human sense-making and communication

Enterprise modeling within category I is the main focus of this thesis: How to use enterprise modeling to support human sense-making and communication in organizations. Enterprise modeling within category I focuses on the modeling process -- how to ensure that an appropriate perspective is constructed and understood, and how to enable perspective taking by other communities.

A model developed as a part of category I enterprise modeling may be "anything" as long as it fulfills the requirement from section that it is a representation of something else, and that the UoD is perceived to be an enterprise. Hence, there are no requirements to medium, formality of modeling language or use of technology.

Category II: Computer-assisted analysis

The main objective within category II is to develop models that can be used to produce answers to questions. Using computers, one may reduce the burden of "manual" execution of models as is required if models from category I are to be used. Computers may efficiently calculate critical paths in project plans, summarize cost figures, simulate effects of coordination policies, etc. Christiansen's work on computational organizational models is an example within category II (Christiansen, 1993).

The main focus is on enterprise models per se, in order to make them accurate enough to match the UoD. Modeling as an activity is not necessarily important in itself as long as the model has the desired qualities. An observation is that one do not "create" anything new from interaction with the model as much as retrieve information (simulation may be considered retrieval of information in a broad sense, although simulation may visualize dynamic behavior or find patterns in data not practically feasible for humans to see without computerized support).

Category II represents the main objective of creating models within the traditional engineering disciplines: Construction of models of artifacts (product models) in order to test them before building them (Rumbaugh et al., 1991:15).

Category III: Model deployment and activation

Within category III of enterprise modeling, the model becomes a part of the actual production system or information system in the organization. The model is explicit, as opposed to the way models are used in information systems development, where conceptual models are used initially to gain understanding, but the final information system is independent of the model (changes to the model is not automatically reflected in the information system). Model deployment and activation corresponds to process model enactment within the software process modeling area (Finkelstein et al., 1994) and to workflow automation within the workflow management domain (Marshak, 1994).

The essence of enterprise modeling within category III is to change the world more directly than in category II: In category III, one may think of the enterprise models actually employed in doing work. The enterprise model and the associated computerized interpreter may be considered an active actor in the organization.

3.5.2 Purposes observed in the empirical studies

The main purpose in common to all three projects was to improve the understanding of Statoil business. Modeling in the VPT and PA30 projects clearly belongs to category I. However, it was the modeling teams that needed to improve their understanding of the business in order to suggest new projects or improvements to the processing plant.

Modeling in the Gazz project was originally intended as a way to think more holistically -- as a tool to support strategic thinking. In this sense, one may consider enterprise modeling in Gazz to belong to category I. The purpose was not to compute answers. However, as the customers shifted to also incorporate engineers, the purpose of modeling was to come up with more realistic figures for capacities and economy. Hence, the modeling effort is categorized as of type II in the taxonomy.