CAJ | 학술논문

为了对动态变化的决策表进行高效属性约简处理,在改进的分辨矩阵的基础上提出一种基于约简树的增量式属性约简算法IRART,该算法首先根据序贯属性约简算法对原决策表构造约简树,然后求出新增对象的分辨向量,并利用此向量对约简树进行修整,从而快速得到新决策表的所有约简,最后通过示例证明了这种算法的有效性.与传统增量式属性约简算法相比,该算法避免了复杂的逻辑演算,提高了属性约简的更新效率,理论分析表明该算法是有效可行的.
위료대동태변화적결책표진행고효속성약간처리,재개진적분변구진적기출상제출일충기우약간수적증량식속성약간산법IRART,해산법수선근거서관속성약간산법대원결책표구조약간수,연후구출신증대상적분변향량,병이용차향량대약간수진행수정,종이쾌속득도신결책표적소유약간,최후통과시예증명료저충산법적유효성.여전통증량식속성약간산법상비,해산법피면료복잡적라집연산,제고료속성약간적경신효솔,이론분석표명해산법시유효가행적.
For the efficient attribute reduction of dynamic decision table,the incremental algorithm for attribute reduction based on discernibility matrix and reduction tree is proposed.This method builds reduction tree according to sequential attribute reduction algorithm,calculates discernibility vector of new object,and revises reduction tree according to discernibility vector.Thereby attribute reduction cluster of new decision table can be abtained quickly,finally the validity of the algorithm is proved by exam-ples.Compared with the traditional algorithm,this algorithm avoids complex logical calculus and improves the updating efficiency of attribute reduction.Theoretical analysis shows that the algorithm of this paper is efficient and feasible.