The Community for Technology Leaders
Green Image
Representation of uncertain knowledge using a Bayesian network requires acquisition of a conditional probability table (CPT) for each variable. The CPT can be acquired by data mining or elicitation. When data are insufficient to support mining, causal modeling, such as the noisy-OR, aids elicitation by reducing the number of probability parameters to be acquired from human experts. Multiple causes can reinforce each other in producing the effect or can undermine the impact of each other. Most existing causal models do not consider causal interactions from the perspective of reinforcement or undermining. Our analysis shows that none can represent both interactions. Except the RNOR, other models also limit parameters to probabilities of single-cause events. We present the first causal model, the non-impeding noisy-AND tree, that allows encoding of both reinforcement and undermining. The model generalizes several existing models for the binary case. It supports efficient CPT acquisition by elicitating a partial ordering of causes in terms of a tree topology, plus necessary numerical parameters. It also allows incorporation of probabilities for multi-cause events.
probabilistic reasoning, Uncertainty, knowledge acquisition, knowledge modeling, elicitaion methods

Y. Xiang and N. Jia, "Modeling Causal Reinforcement and Undermining for Efficient CPT Elicitation," in IEEE Transactions on Knowledge & Data Engineering, vol. 19, no. , pp. 1708-1718, 2007.
82 ms
(Ver 3.3 (11022016))