A Methodological Approach for the Effective Modeling of Bayesian Networks
Modeling Bayesian networks manually is often a tedious task. This paper presents a methodological view onto the effective modeling of Bayesian networks. It features intuitive techniques that are especially suited for inexperienced users: We propose a process model for the modeling task, and discuss strategies for acquiring the network structure. Furthermore, we describe techniques for a simplified construction of the conditional probability tables using constraints and a novel extension of the Ranked-Nodes approach. The effectiveness and benefit of the presented approach is demonstrated by three case studies.
Unable to display preview. Download preview PDF.
- 1.Russell, S., Norvig, S.: Artificial Intelligence: A Modern Approach, 2nd edn. Prentice–Hall, Englewood Cliffs (2003)Google Scholar
- 2.Wrobel, S.: An Algorithm for Multi-Relational Discovery of Subgroups. In: Proc. 1st Europ. Symp. Principles of Data Mining and Knowledge Discovery, pp. 78–87. Springer, Berlin (1997)Google Scholar
- 5.Koller, D., Pfeffer, A.: Object–Oriented Bayesian Networks. In: Proceedings of the Thirteenth Conference on Uncertainty in Artificial Intelligence (UAI 1997), pp. 302–313 (1997)Google Scholar
- 6.Neil, M., Fenton, N., Nielsen, L.: Building Large-Scale Bayesian Networks. Knowledge Engineering Review (1999)Google Scholar
- 7.Fenton, N., Neil, M.: Ranked Nodes: A Simple and Effective Way to Model Qualitative Judgements in Large–Scale Bayesian Nets. IEEE Transactions on Knowledge and Data Engineering (2005)Google Scholar
- 9.Helsper, E., van der Gaag, L., Groenendaal, F.: Designing a Procedure for the Acquisition of Probability Constraints for Bayesian Networks. In: Motta, E., Shadbolt, N.R., Stutt, A., Gibbins, N. (eds.) EKAW 2004. LNCS (LNAI), vol. 3257, pp. 280–292. Springer, Heidelberg (2004)Google Scholar