An Improved Aggregated One-Dependence Estimator:On Not So Rigid Cross Validation，郑庆华，，The naive Bayesian classifier is a simple and effective approach to classifier learning. However, its conditionally independent assumption doesn’t often hold in the real world