Model generators: Prototyping simulation model definition, specification, and documentation under the conical methodology

Abstract

The process of model generation is key to the realization of a Simulation Model Development Environment. Model generation is facilitated in the environment via the Model Generator - a software utility that assists a modeler in the development of a simulation model specification. Since modeling is inherently creative, the correct assistance provided to a modeler can neither be derived algorithmically, nor proved mathematically. Only through experimentation with prototypical assistance forms can we begin to understand the meaning of correctness.

This thesis describes the development of a Model Generator prototype for the Simulation Model Development Environment. A review of the literature indicates the need for more extensive questioning of the model generation process to identify the proper foundational support than 1s available in applications designed under the program generation approach. The Conical Methodology provides the Conceptual Framework, and the Condition Specification provides the target specification form for the Model Generator prototype.

A set of algorithms to derive a condition specification via a series of interactive dialogues is presented, and the results of early prototype experimentation are discussed. New questions are raised as to the role of relational attributes in the Conical Methodology and the extent and types of model analysis provided by a Model Generator. Finally, an analysis of the Model Generator as a platform for the assessment of the Conical Methodology/Condition Specification is given and directions for future research outlined.