Crossing the Chasm – From Business Models to Software Architecture

The paper presents an original agent-oriented methodology suitable for development process of management information systems for virtual organizations. It starts with description of emerging business structures and a semantic gap problem, which is quite common for software development projects supporting operations of modern enterprises. To eliminate this problem we have decided that system architecture source code will be generated directly from business models describing virtual organization strategy, structure, business processes, business rules and assets. The business models will be created with specific modeling methods and visual language created with UML extensions we have developed for our methodology. Using software agent and its basic characteristics as main implementation construct enabled us to provide close semantic proximity between business models and system’s architecture components. Therefore it was possible to create the transformation templates that can be run in a CASE Tool environment (Enterprise Architect) with MDG Technology to generate, directly from visual models, runnable source code of the software solution that possesses high level of business value. In our opinion it is significant contribution because we have reduced the Business–IT Wall which is predominant in management information systems development projects. All modeling methods included in our approach have been based on international standards published by Object Management Group – Business Motivation Model (BMM), Business Process Model and Notation (BPMN), Unified Modeling Language (UML) and Software Process Engineering Metamodel (SPEM). This should shorten the adoption timeframe of our methodology when used by teams that are familiar with state-of-the-art business computing standards. For implementation discipline in our methodology we have adopted Belief-Desire-Intension architecture for Multi-Agent Systems that is widely used framework and its implementation – Jason environment with Agent Speak language. Before we have decided to develop brand new methodology, comparative analysis of several currently used approaches aimed at inter-organizational information systems modeling has been conducted. The analysis results, which summary is presented in the paper, have enabled us to formulate the motivations and assumptions for the approach taken. The paper also discusses the static and dynamic structure of the methodology and the example of its possible use case.