摘要
属性约简是粗糙集研究的重要内容之一。目前有多种计算约简集的方法,但计算效率普遍不高。杨萍等学者提出的基于二进制区分矩阵的启发式约简算法,考虑了属性的区分度和区分率,采用高效的逻辑运算获得约简集,提高了运算的效率。在该算法的基础上,首先指出其计算所得的约简集存在不确定性,然后给出一种考虑属性排名的改进的约简算法,消除了约简集的不确定性,并且可以迎合用户的需求。最后通过一个信息系统实例,验证该算法的可行性和有效性。
The attribute reduction is one of the major contents of rough set research.There are a variety of methods based on rough set theory to compute reduct set of an information system,but their computational efficiency is always not high.Yang Ping proposed a heuristic reduction algorithm based on binary-valued discernibility matrix,taking into account the degree and ratio of differentiation between condition attributes,adopting efficient logical operator computation to obtain reduct set.Based on the thesis of Yang Ping＇s algorithm,first of all points out the result of Yang Ping＇s algorithm is a reduct with some kind of uncertainty,then proposed an improved algorithm considering ranking of condition attributes.It eliminates the uncertainty of final reduct and meets the needs of user simultaneously.Finally,an information system example is used to show its feasibility and effectiveness.
出处
《计算机技术与发展》
2010年第12期82-85,共4页
Computer Technology and Development
基金
国家自然科学基金(10771171)
兰州市科技计划项目(2008-1-34)
关键词
属性排名
粗糙集
属性约简
二进制区分矩阵
ranking order of attributes
rough set
attribute reduction
binary discernibility matrix
