11.1    Research Questions Revisited

The findings reported in the thesis are summarized by returning to the research questions and objectives posed in section 1.2.2, and recapitulating the work that has been done to answer them. Selected results are highlighted.
Main Research Question 
How is and should enterprise modeling be performed when the main purpose of modeling is to support human sense-making and communication?

The main research question has two parts: An is part concerning enterprise modeling practice (a descriptive account) and a should part concerning a possible approach to modeling (a prescriptive account). The approach followed to answer the two parts were formulated as two research objectives RO1 and RO2, respectively:

Research Objectives 
RO1: Based on empirical studies of real enterprise modeling projects, formulate a set of assertions concerning enterprise modeling practice when the purpose of modeling is human sense-making and communication. 

RO2: Develop a methodological framework based on the above-mentioned assertions to guide practitioners in their enterprise modeling efforts.

The empirical studies conducted to meet RO1 have been the strategy project and to some degree the three initial projects VPT, PA30 and Gazz. RO2 has been sought met by development of Tema a framework for enterprise modeling that subsumes the principles developed to meet RO1 and thereby ensures a grounding of the framework in observations of enterprise modeling practice.

The main research question was elaborated into four partial questions RQ1 - RQ4 in order to reduce the number of aspects of enterprise modeling that were to be investigated. Each question is discussed separately below.

RQ1: What are the purposes of developing and using enterprise models, and how are enterprise models actually used? 

The purposes of enterprise modeling have been discussed both as a part of empirical studies in chapters 2 and 6 and in the state-of-the-art survey in chapter 5. A partitioning into three categories of enterprise modeling based on purpose was introduced in section 3.5, with human sense-making and communication as one of the categories (and the focus of this research project). Most of the objectives observed in the empirical studies were within the confines of this category (referred to as category I).

The most distinguished use of models observed in the empirical studies (and stated as assertion 16) was as structuring devices: The models were used to decompose the domain and then focus on the elements one by one (reducing complexity) or the relationships between the elements (holistic thinking).

RQ2: What activities are most important in an enterprise modeling process? 

The enterprise modeling process was investigated in terms of the empirical studies, general theories and the state-of-the-art survey.

In the empirical studies, enterprise modeling processes were observed to depend upon how well known the domain was to the modeling actors. The type of work that was modeled seemed to influence how well the domain was known. If the domain was dominated by routine work and material artifacts, modeling was dominated by representation. When the type of work was less known (e.g., being non-routine work and information-based artifacts), modeling became more an act of sense-making. When modeling was dominated by sense-making, the modeling process became highly unstructured and arbitrary, as discussed in section 6.6.2.

In the contemporary frameworks discussed as a part of the state-of-the-art survey in chapter 5, modeling was mainly presented as a structured method. A few exceptions were found in F3, SSM and OOram, relying on more informal guidelines and driving questions. The stance taken on activities comprising enterprise modeling in Tema was as an unstructured, opportunistic, creative and analytical process using a pool of techniques.

RQ3: What are the most important properties of enterprise models, i.e., the artifacts or manifestations of the modeling process? 

Based on observations from the empirical studies, artifacts were divided into two categories primary artifacts and subsidiary artifacts. Enterprise models, being primary artifacts, were observed to be simple (in the sense that there were few language concepts and few elements in the models), informal (in the sense that the modeling languages had no formally defined syntax or semantics) and tailored to the specific needs of the projects studied (languages were developed as a part of the modeling process).

Concerning subsidiary artifacts (serving a purpose in the modeling process but not considered as enterprise models per se), terminology, model explanations, metaphors and narratives were observed to play a role in the empirical studies (and most notably in the strategy project of chapter 6). How these could be integrated and used in conjunction with enterprise models was indicated in Tema.

RQ4: What are the most prominent relationships between purposes, modeling processes, models, modelers and the enterprise that is modeled? 

Relationships between various aspects of enterprise modeling were proposed in terms of the principles outlined in chapter 7, and mainly concerned the knowledge and competence of both the modeling actors and the enterprise audience related to the Universe of Production or the Universe of Modeling.