论文部分内容阅读
随着形式背景中数据的增多,概念数量会急剧增加。基于决策形式背景的属性约简在保持决策规则分类能力不变的前提下,寻找极小属性子集,使得决策规则得以简化。文章首先将规则分为强规则与弱规则,提出非冗余规则的判定定理及规则约简的判定定理并予以证明;其次提出规则约简及规则输出算法,具体做法是:生成非冗余规则,然后对非冗余规则进行约简,保留规则中相对必要属性的最简形式,删除规则中的不必要属性;随后讨论了算法的时间复杂度。通过实例分析,对比了其他属性约简算法的运行效率和分类能力,证明本文提出的算法具有可行性和正确性。
As the amount of data in the formal context increases, the number of concepts increases dramatically. Based on the decision-making formal context attribute reduction under the premise of unchanged decision-making classification ability, to find the smallest subset of attributes, making the decision-making rules can be simplified. Firstly, the rules are divided into strong rules and weak rules, and the judgment theorems of non-redundant rules and rules of rules reduction are proposed and proved. Secondly, rule reduction and rule output algorithms are proposed. The rules are as follows: Generating non-redundant rules , Then reduce the non-redundant rules, keep the simplest form of the relative necessary attributes in the rules, delete the unnecessary attributes in the rules, and then discuss the time complexity of the algorithm. Through the example analysis, compared with other attribute reduction algorithms running efficiency and classification ability, this paper proves that the proposed algorithm is feasible and correct.