关键词: 多关系数据挖掘, 朴素贝叶斯, 参数平滑, 互信息

Abstract:  To eliminate the naive Bayesian classification of zero probability and overfitting problem, this paper discusses the various probability smoothing method, gives MRNBC-M（Multi-Relational Naive Bayesian Classifier based  on M-estimation）and MRNBC-L（Multi-Relational Naive Bayesian Classifier based  on Laplace-estimation）. In the case of multi-relationship, the impact of M and Laplace estimation methods on the classification is analyzed. In order to further optimize the classification, the method is based on the extended mutual information criterion. Experiments on the multi-relational datasets show that MRNBC-M can effectively improve the classification performance.

Key words:  Multi-Relational Data Mining（MRDM）, Naive Bayes, smoothing methods, mutual information