Traditional and Object Oriented approaches to software development have models for the development life cycle of large software systems. Even when not followed in detail, life cycle models provide useful guidance to a complex task. The development of large scale Bayesian nets differs from other types of software development. These differences are a result of the Knowledge Engineering process needed for Bayesian Net development. Hence the lack of a software development approach that allows for Knowledge Engineering has had a significant impact on the uptake of Bayesian net technology.
This paper discusses some Knowledge Engineering approaches for Bayesian nets and presents the KEBN life cycle development model which accommodates the use of Knowledge Engineering and is specific to the domain of large scale Bayesian net development. The KEBN approach is compared to other traditional life cycle models and shown to be more compatible with large scale Bayesian net development.
Read the full paper.
Andre Oboler, The KEBN Process: A new approach to Knowledge Engineering with Bayesian Nets, School of CSSE, Monash University, 2002