海量数据且高维环境下,朴素贝叶斯分类可能即面临获取大量带类标签代价过高又面临当前分类规则不能适应数据变化等问题。于是提出一种基于小规模训练集的基于粗糙集(RS)动态约简增量式贝叶斯算法来实现问题分类:利用粗糙集理论对数据进行动态属性约简获得极小集属性,再用朴素贝叶斯推理(NBC)方法对知识进行增量学习和分类。该方法结合了增量学习与动态约简的优点。实验证明了算法的可行性。 Under the circumstance of massive data and high dimensional environment, Naïve Bayesian classification may face such problems as too much price tagging with tags and the current classification rules can not adapt to data changes. Therefore, an incremental Bayesian algorithm based on dynamic reduction of rough set (RS) based on small-scale training set is proposed to realize the classification of the problem: the rough set theory is used to reduce the dynamic attribute of the data to obtain the minimal set of attributes, The Bayesian Inference (NBC) approach to incremental learning and classification of knowledge. This method combines the advantages of incremental learning with dynamic reduction. Experimental results show that the algorithm is feasible.