摘要
分析HORAFA算法和HORAFA-A算法的不足,给出一种获得最优约简的启发式算法。算法以核属性为初始约简集,以属性频率为启发式信息,选择必要的属性加入约简集。该算法不仅适用于相容决策表系统,也适用于不相容决策表系统;同时,改进了反向消除方法,可以更快速地删除多余条件属性。实验表明,该算法是正确的,并且效率优于HORAFA-A算法。
Analysis of the disadvantages of HORAFA and HORAFA-A is made and an optimized reduction algorithm is put forward. The algorithm uses core attributes as the initial sets ,uses attribute frequency as the heuristic information, and selects necessary attributes to add to the reduction set. The algorithm suites both the compatible decision table system and the incompatible decision table system. Meanwhile, the improvement of the converse elimination can delete the surplus condition attributes more quickly. The experimental result show that the algorithm is correct and is better than HORAFA-A.
出处
《计算机应用与软件》
CSCD
2009年第3期11-13,17,共4页
Computer Applications and Software
基金
国家自然科学基金项目(60273043)
安微省自然科学基金(O504202204)
安微高校省级自然科学研究项目(KJ2007B237
KJ2008B117)
关键词
粗糙集
可分辨矩阵
核属性
属性频率
Rough set Discernibility matrix Core attributes Attribute frequency
