In this thesis, I tackle several open problems in Bayesian networks by drawing from corresponding techniques already established on relational databases. More specifically, I focus on three important tasks: (i) the incorporation of evidence into a Bayesian network, (ii) the modelling of multiagent Bayesian networks, and (iii) the use of index structures when implementing probabilistic expert systems. I explicitly illustrate the many advantages of our proposed methods. Thereby, the work in this thesis not only addresses three significant problems in Bayesian networks, but it also further emphasizes the intrinsic relationship between Bayesian networks and relational databases.
展开▼
