关键词: 属性约简, 正区域, 核, 粗糙集

Abstract: The efficiency of attribute reduction is a key issue in rough set and other soft computing theories.In order to enhance it，a new algorithm for computing core is proposed after analyzing the indiscernibility relation and radix sorting，and its time complexity is O（|C||U|）.Furthermore，a quick reduction algorithm which uses improved attribute significance as heuristic information is presented，the time complexity is O（|C|²|U|）.Finally，through some experiments on the data sets in UCI machine learning repository，the algorithm is proved more efficient and suitable for large data sets.

Key words: attribute reduction, positive region, core, rough set