This paper proposes an idea of reverse modeling that analyses LP models and then converts them into an object-oriented model repository. The process of reverse modeling consists of (1) identifying and analyzing source models by meta processor, (2) model decomposition and generalization to scan the models and divide them into model components, and (3) deriving model selection rules from the components by rule generator. Through the process, the authors can extract reusable model components and build a model base with model selection rules. Examples with models created by SML and MODLER modeling languages are given to illustrate the methods. The model base maangement capabilities provided by reverse modeling can increase the reusability of current modeling tools. [In Korean.]
