for many applications is the ability to induce models from
data. This ability allows to complement expert knowledge
with data to improve performance of a system.
In the last decade there has been an active research effort to
develop the theory and algorithms for learning of Bayesian
networks from data. This includes methods for parameter
learning [1, 19, 27] and structure learning [3, 6, 16, 26].
Using structure learning procedures we can learn about the
structure of interactions between variables in an unknown
domain.