Abstract: Fuzzy association rules have been used to deal with imprecision in databases and offer a good representation of knowledge discovery. Using restriction level representation theory, GUHA model is generalized for fuzzy association rules. By restriction levels managing fuzzy rules, an extended validation measurement process is presented. Mining algorithm using the formal method, parallels the mining process in different restriction levels, and then summarizes the obtained results. Algorithm complexity analysis and experimental results show that this method is feasible and effective, which sets a logic basis for the representation and evaluation of fuzzy association rules.

Key words: fuzzy association rules, formal method, Restriction Level（RL）, General Unary Hypothesis Automaton（GUHA） model, support, confidence, certainty factor

摘要： 模糊关联规则用于处理数据库中的不精确信息，并提供一个知识发现的良好表示。利用约束级别表示理论将GUHA模型泛化用于模糊关联规则，通过约束级别管理模糊规则，并给出一个扩展的验证度量过程。使用形式化方法的挖掘算法，在不同的约束级别上并行化挖掘过程，总结得到的结果。算法的复杂度分析以及实验结果表明该形式化方法是有效可行的，从而确立了模糊关联规则表示和评价的逻辑基础。

关键词: 模糊关联规则, 形式化方法, 约束级别（RL）, 一般一元假设自动机（GUHA）模型, 支持度, 置信度, 确定性因子