Research on Bayesian Network Structure Learning Based on Rough Set

Abstract

Rough set theory and method is one kind of effective method for dealing with complicated system, but it fails to contain the theory and mechanism handling imprecise or uncertain data. So, it has strong complementarities with Bayesian network theory. The paper puts forward a kind of Bayesian network structure learning method combining rough set theory with Bayesian network. Inclusion theory of rough set is used to mine cause and effect associated rules which determine arc and its direction between Bayesian network variables. At the same time, mining arithmetic of associated rules is presented in the paper. Finally, it shows rationality and validity of the approach through experiment analysis.