基于改进分辨矩阵的增量式数据挖掘模型研究

粗糙集理论是近年来出现的处理模糊和不确定性的数学工具,并已广泛应用于人工智能的许多领域。文章针对在增量式数据环境下挖掘决策规则的特点,在回顾基于分辨矩阵的数据挖掘算法及其不足的基础上,利用决策矩阵的概念有效地处理具有不同决策类的各种决策系统。在此基础上提出相应的对每一个决策类建立决策矩阵的增量式挖掘算法,最后利用算例验证了算法的合理性和有效性。该算法步骤同传统的分辨矩阵算法相比,能在增量式环境下快速而有效地进行确定性规则和可能性规则的学习并对可能性规则建立相应的置信度,使规则的获取更具实用性。

Research on Incremental Data Mining Model Based on Improved Discernibility Matrix

Rough set theory has recently become popular mathematical tool to solve problems on fuzzy and uncertainty used in the fields of artificial intelligent.According to the characteristic of decision rules mined in the environment of incremental data set and on the base of the limitation of algorithm of discernibility matrix,this paper makes use of the concept of decision matrix to deal with all varieties of decision system with different decision type.After that incremental data mining algorithm of establishing decision matrix to each of decision type is put forward and the characteristic of rationality and validity are examined by example.Compared with the traditional algorithm of discernibility matrix,this incremental algorithm can effectively produce deterministic rules and probabilistic rules with corresponding probability,so the rules mined by this algorithm has more practicability.