Olav Bangsø, Helge Langseth, and Thomas D. Nielsen, Aalborg University, Denmark
When constructing a Bayesian network, it can be advantageous to employ structural learning algorithms to combine knowledge captured in databases with prior information provided by domain experts. Unfortunately, conventional algorithms do not exploit the occurrence of repetitive structures, which are often found in object oriented domains such as fault prediction in computer networks and large pedigrees. In this paper we propose a method for structural learning in object oriented domains. It is demonstrated that this method is more efficient than conventional algorithms, and it is argued that the method supports a natural approach for expressing the prior information of domain experts.