基于数据相容填补的极大相容块构造算法  

Algorithm of Constructing Maximal Consistent Block Based on Consistent Data Reinforcement

作　　者：周石泉[1] 蒙祖强[1] 

机构地区：[1]广西大学计算机与电子信息学院,南宁530004

出　　处：《计算机科学》2012年第9期192-197,共6页Computer Science

基　　金：国家自然科学基金项目(61063032);广西教育厅科研基金项目(201012MS010)资助

摘　　要：极大相容块技术在不完备信息处理方面具有一定优势,但极大相容块获取本身是个耗时的过程。通过分析不完备信息系统中极大相容块的性质特点,给出极大相容块中不完备数据的相容填补方法;在不完备信息系统中使用相容填补方法,得到一种新的极大相容块构造算法;基于构造算法的特点,引入决策树存储结构对算法进行优化。使用标准的数据集验证了构造算法的有效性。实验结果表明,构造算法在较大规模的数据集上有较好的性能。Maximal consistent block technique has some advantages in dealing with the incomplete information.But the construction of maximal consistent blocks itself can be a time-consuming process.After the properties of maximal consistent blocks in incomplete information system were analyzed,the consistent data reinforcement method was defined to handle the missing data in maximal consistent blocks.Using the method in incomplete information system,a new algorithm was obtained to construct maximal consistent block.And the decision tree structure was introduced to optimize the algorithm based on the algorithm’s characteristics.Algorithm was evaluated on the standard benchmark dataset.The result of the experiments indicates that the algorithm has better performance in large scale dataset.

关 键 词：粗糙集 不完备信息系统 极大相容块 数据填补 决策树 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]