The Reduction Algorithm for Dynamic Information Systems Based on Distinguishable Relation
To capture a dynamic system with static reduction algorithm, the cost will enlarge a lot for the repeated work. Moreover, the algorithm can not catch partial changes in the system. To smooth this issue, a dynamic attribute reduction algorithm is introduced when the universe of the system keeps the same and the condition attribute set is added. The concept, Flagged Matrix, K-Flagged Matrix and ∑-Flagged Vector is defined and a reduction algorithm for dynamic information systems is given based on distinguishable relation. This algorithm will use the new added attributes and corresponding matrixes to correct and adjust the original attribute reduction. To do the reduction for the dynamic systems, as the theoretical illustration and empirical analysis turns ont in this paper; the new added attributes are compared with the original attributes considering the distinguishable ability using corresponding matrixes. Then evaluate the added attributes are the core attributes or not, and determine whether the added attributes can replace the original attributes or not. After that, a new reduction is got. This algorithm can figure out the dynamic attribute reduction efficiently. For it will use matrix operation instead of logical operation, the cost of this algorithm is significantly decreased.
data mining rough set dynamic reduction distinguishable relation
Xinying Chen Kesheng Xu
School of Software Technology Dalian Jiao Tong University Dalian China
国际会议
The 2nd IEEE International Conference on Advanced Computer Control(第二届先进计算机控制国际会议 ICACC 2010)
沈阳
英文
60-64
2010-03-27（万方平台首次上网日期，不代表论文的发表时间）