A MODEL INTEGRATION FRAMEWORK FOR MODELING IN THE LARGE

Abstract

Model integration is ever more important in practice as a result of several industry trends, toward multi-company cooperative arrangement up and down the supply chain, and toward finding new uses for expensive assets. From an organizational view, model integration is necessary at the top management level where more complex strategic decisions and broader sensitivity analysis are in demand. Some progress has been made in model integration. These various approaches have contributed to the different stages within the overall modcling life cycle. Model integration is a particularly crucial operation which requires thinking about "modeling in the large", and extends the scope of model management research to include manipulation as well as definition. For modeling in the large, one needs to consider what process integration should be applied for an integrated schema. In this thesis. We present a model integration framework based on Object-Oriented Model Representation (OOMR). The framework establishes an inference mechanism to link schema integration and process integration. Operations in schema integration will be transformed to corresponding actions in process integration. Model integration can therefore be implemented in the overall modeling life cycle. We describe the framework details as follows. First, the internal model representation with object-oriented properties, called Object-Oriented Model Representation (OOMR) is discussed. In OOMR, a model is viewed as a collection of model pieces and is represented by multiple abstractions which form a model class hierarchy of model type, model template and model instance. aggregation and specialization arc two important features of OOMR The aggregation structure allows the combination of model pieces in the model framework, whereas specialization structure builds an inheritance mechanism between the different abstractions of a model. When two model schemas, which correspond to model template classes, are integrated, corresponding model types will be determined through inheritance of the model class hierarchy. Second, we introduce operations of schema integration. By adopting model schema of structured modeling (SM), a structured model schema in OOMR is formed. The structured model schema in OOMR as a user view reflects the structure of an internal model class. The structured model schema in OOMR is represented in the genus graph. Model schema integration is carried out on the genus graph. Third, integrated rules to derive process integration from schema integration are proposed. A group of integrated rules is used to identify variable correspondence at every integrated joint and lo derive synchronization of individual processes. Based upon the complexity and dynamism of variable correspondence and synchronization, typical types or process integration are identified. Finally, the relationship between schema integration and process integration is discussed. The discussion shows how the integration framework works for modeling in the large.