Decision Aid for Selecting Representation Strategy

The ideal set of materials we would like to wind up with would include the following:

1. Requirements for domain modeling representations
2. A set of alternative representative strategies
3. Assessment criteria; characteristics and attributes that would help decision-makers select appropriate representations;
4. A feature model placing the example strategies within an overall comparative framework, structured in terms of differentiating features;
5. Mapping of assessment criteria to alternatives, to form a decision aid for domain modeling project planners;
6. A set of case studies, characterized in terms of the attribute profiles called out in the assessment criteria, the representation strategies selected, and the rationale for the selection.
7. Lessons learned from the process of developing the framework.

This overall framework could be used in a variety of ways. The primary audience would be domain modeling project planners, who would use the framework as an assessment and educational tool. The assessment criteria would help identify relevant issues to be considered in selecting an appropriate domain model representation strategy. The comparative framework would educate planners, and later modelers, about the range of alternatives available, their strengths and weaknesses, and would suggest strategies for addressing the risks and avoiding the pitfalls of various notations.

The framework would also be useful as new representation strategies are identified or developed, and/or as tool support for domain modeling matures. For example, CASE tool vendors might use the framework to determine ways of extending system representations to support semantically clean conventions for modeling commonality and variability. Methodology developers could either characterize their chosen representation strategies in terms of the framework, or could reference the framework itself as a decision aid at the appropriate point(s) in the domain analysis process they describe.

Finally, the framework (together with lessons learned from the process of solidifying the framework) could serve as a template for other supporting methods of the domain analysis discipline. In the long term, as the gaps are filled in, we could imagine a suite of documents or support mechanisms that would help project planners configure the best set of methods, processes and representations for a domain engineering project within their organization.