CAJ | 학술논문

粗糙集中找到最小属性约简是一个NP-hard问题，本文根据知识信息熵的定义和性质，定义属性重要性，分析其性质，得出核的求法，给出约简的判定定理，提出了一种基于信息熵的属性约简算法。该算法采用启发式搜索法，先计算属性重要性，求得核，再以核为起点，以属性重要性大小为启发式信息，选择属性，求得最小约简。理论分析和实际计算表明，该算法简明有效。
조조집중조도최소속성약간시일개NP-hard문제，본문근거지식신식적적정의화성질，정의속성중요성，분석기성질，득출핵적구법，급출약간적판정정리，제출료일충기우신식적적속성약간산법。해산법채용계발식수색법，선계산속성중요성，구득핵，재이핵위기점，이속성중요성대소위계발식신식，선택속성，구득최소약간。이론분석화실제계산표명，해산법간명유효。
Finding minimum attributes reduction is a NP-hard problem in rough set. This paper defined importance of attributes based on definition and properties of knowledge entropy, made the analysis of its nature, gave the solution of core and the reduction theorem and proposed an algorithm of attribute reduction based on entropy. The algorithm uses heuristic search method. First it cal-culates the attribute importance and gets core, then taking the core as a starting point and making size of attribute importance as heu-ristic information, it selects attributes and gets the minimum attributes reduction. The theoretical analysis and example calculation show the algorithm is simple and effective.