Metamodeling of Bayesian networks for decision-support systems development

The knowledge modeling and software modeling phases in Knowledge-Based System development are not integrable, in terms of representation, due to the different languages needed at the steps of the development. This paper focuses on bring closer these languages. By one hand, we define a meta model which contains the key concepts used in the definition of a knowledge model as a Bayesian network. On the other hand, we define an extension of UML using profiles that can bridge the gap in representation and facilitate the seamless incorporation of a knowledge model, as Bayesian network, in the context of a knowledge-based software development