Study of MEBN Learning for Relational Model.
Cheol Young Park, Kathryn Blackmond Laskey, Paulo Costa, Shou Matsumoto · 2012
Abstract — In the past decade, Statistical Relational Learning (SRL) has emerged as a new branch of machine learning for representing and learning a joint probability distribution over relational data. Relational representations have the necessary expressive power for important real-world problems, but until recently have not supported uncertainty. Statistical relational models fill this gap. Among the languages recently developed for statistical relational representations is Multi-Entity Bayesian Networks (MEBN). MEBN is the logical basis for Probabilistic OWL (PR-OWL), a language for uncertainty reasoning in the Semantic Web. However, until now there has been no implementation of MEBN learning. This paper describes the first implementation of MEBN learning. The algorithm learns a MEBN theory for a domain from data stored in a relational Database. Several issues are addressed such as aggregating influences, optimization problem, and so on. In this paper, as our contributions, we will provide a MEBN-RM (Relational Model) Model which is a bridge between MEBN and RM, and suggest a basic structure learning algorithm for MEBN. And the method was applied to a test case of a maritime domain in order to prove our basic method.