Designing a pattern language to enhance model composability and reusability: An example with component-based probabilistic models

This paper presents a pattern language for developing Object-Oriented Bayesian Networks (OOBN), as a member of the component-based probabilistic models family, to tackle complex problems. The proposed pattern language integrates knowledge from various domains, such as modeling, software engineering, and Bayesian networks, to provide a comprehensive solution for developing OOBNs. The paper also provides a validation framework to evaluate the pattern language. As a practical application for the OOBN pattern language, a case study of using it to develop an OOBN is presented. The model in the case study aims to represent the complex interconnections among the Sustainable Development Goals (SDG), long-term sustainability and resilience. The results of the case study validate the effectiveness of the pattern language and highlight its potential for future applications. The proposed OOBN pattern language provides a systematic approach to the development of OOBN, reducing the complexity and increasing the efficiency of their modeling process