朴素贝叶斯分类器不能有效地利用属性之间的依赖信息,而目前所进行的依赖扩展更强调效率,使扩展后分类器的分类准确性还有待提高.针对以上问题,在使用具有平滑参数的高斯核函数估计属性密度的基础上,结合分类器的分类准确性标准和属性父结点的贪婪选择,进行朴素贝叶斯分类器的网络依赖扩展.使用UCI中的连续属性分类数据进行实验,结果显示网络依赖扩展后的分类器具有良好的分类准确性. Naïve Bayesian classifier can not effectively use the dependency information between attributes, and the current dependence expansion emphasizes more efficiency, so the classification accuracy of the extended classifier needs to be improved.In view of the above problems, Based on Gaussian kernel function estimation of attribute density, the network dependent expansion of naive Bayesian classifier is combined with the classification accuracy criteria of classifiers and the greedy selection of attribute parent nodes. The results show that the network-dependent classifier has good classification accuracy.