Model Oriented Composition Theory

With the rapid development of computer science and software engineering, there is a growing awareness of the important influence on software analysis and designing made by formal models. Many modeling methodologies and tools are employed to model the real-world scenarios when designing and developing modern software systems. The models are also used to analyze and improve the software systems. It is essential for the designers and developers to have a proper model for a complex software system. Such a model can help them gain a correct view of the system in its development cycle.Since it is difficult to abstract a model covering all required system attributes, the designers usually divide the system according to certain features and establish models for every component with appropriate approaches and tools. This use of divide and concur strategy significantly reduces the complexity of dealing with concrete system at-tributes and facilitates the analysis and reasoning of specific system properties. Modeling methodologies concentrating on distinct system attributes, however, are often developed independently and rely on dissimilar mathematical foundations. We find it hard to share the information from various models and let them cooperate with each other to obtain the expected global view of the system.To solve this problem, a precise and systematic theoretical framework is indispens-able to describe the relationship among these modelisms. The framework also needs to support constructing particular model integration and coordination patterns with informa-tion sharing and interaction among models.The main motivation of this paper is to propose a theory for integrating or coordi-nating different models of software systems. The theory needs to interpret the reasonable model composition patterns; and support the study of the relationship among different models as well as interesting properties of composed models in many practical scenarios. The theory has a capability to describe the system attributes as well as their connections and investigate the relations between distinct behavioral representations. With this theory, we can understand the features of existing models as well as their connections deeply and study the evolution of specific properties during the model composition process. This theory can also serve as a guide to adjust the existing models and combine them to obtain composed models satisfying particular requirements. Furthermore, the theory can con-tribute to achieving unity in perspectives of software designers and software developers, and reducing the nondeterminism of the modeling and development. hence significantly increasing the efficiency of software designing, development and verification. The main contribution of this paper includesWe propose a model representation framework for software systems. The frame-work describes the behaviors of the system based on the observation of its specific attributes. The information of a model covers the relative system attributes and their connections, operations and relations over model instances. We show that the framework can represent many modeling patterns including transformation models and reactive models.We investigate the solutions of composition issues such as program composition and service composition in software engineering field and learn their experience. We capture the general features of composition issues and introduce the concept of model composition. The composition theory applying on models includes model extension, model integration and model completing. It shows how to combine the information between different models and establish composed models with specific properties.