摘要
针对大容量数据表构造的区分矩阵过于庞大致使属性约简算法效率低的问题,引入置信度和支持度,提取大型数据库中的高概率事件,重新构造决策数据表,并在构造区分矩阵过程中剔除重复项和包含项,结果使得比较次数减少、存储空间节省、约简效率提高。
In knowledge system of large database,large discernibility matrix reduces efficiency of attribute reduction algorithm.To solve this problem,confidence and support are introduced to reconstruct the decision table by extracting high probability events of a large database.In the reduction algorithm based on discernibility matrix attribute,the duplicates and contains items are removed to reduce comparison times.As a result,the efficiency of attribute reduction algorithm is improved and the storage space is saved.
出处
《武汉科技大学学报》
CAS
2011年第2期126-130,共5页
Journal of Wuhan University of Science and Technology
基金
国家高技术研究发展计划(863计划)资助课题(2009AA04Z136)
关键词
决策表
区分矩阵
属性频度
属性约简
decision table
discernibility matrix
attribute frequency
attribute reduction