基于归纳的值约简算法的研究与实现

The Study and Implementation of Induction Value Reduction

摘要: 在粗糙集理论的基础上，本文研究了归纳值约简过程。归纳值约简算法采用求解知识表达系统决策表的最小决策算法来求其约简，它可以通过分别求解各个决策规则类的最小决策算法来实现。对于每个决策规则类中的规则，首先计算其核值属性，然后判断核值属性是否能够决定该规则，如果能够决定，则输出规则并删除其等价规则；否则，逐渐加入非核值属性，直到能够决定该规则，然后输出规则并删除其等价规则。最终实现了其测试系统。

Abstract: Based on the research of rough set theory, this paper studies the process of induction value reduction. The induction value reduction algorithm uses the minimum decision algorithm to solve the decision table of the knowledge representation system to obtain the reduction. It can be realized by solving the minimum decision algorithm of each decision rule class. For each rule in each decision rule class, the algorithm computes its core attributes and then determines whether the core attributes can determine the rule. If can, then it will output the rule and remove its equivalent rules. Otherwise, it will gradually add the non-core attributes until they are able to determine the rule, then output the rule and remove its equivalent rules. At last the test system is implemented.