As an effective tool for analyzing human behavior and social cognition, rule extraction is a key issue in cognitive computing. However, the existing association rules state whether an attribute is possessed only without considering the order relation of attribute values. In this paper, we propose a novel method for quantitative association rule mining based on a generalized one-sided context and solid cognitive foundations. The extracted rule indicates the order relation of the attribute values and the structure of truth values for different attributes. We also propose specific algorithms to extract generalized one-sided quantitative association rules and non-redundant generalized one-sided quantitative association rules. The scale of data also needs to be considered by cognitive computing. An object may possess different values for the same attribute according to different measuring scales. The relationship between generalized one-sided quantitative association rules at different scales is also discussed. Rather than converting the multi-valued formal context into a binary formal context, the generalized one-sided quantitative association rule is extracted directly in a multi-valued formal context. The experimental results show the presented algorithms reduce both time and space complexity compared with the classical quantitative association rule-mining algorithm.

作为分析人类行为和社会认知的有效工具，规则提取是认知计算中的关键问题。但是，现有的关联规则规定了是否仅在不考虑属性值的顺序关系的情况下拥有属性。在本文中，我们提出了一种基于广义单侧上下文和扎实的认知基础的定量关联规则挖掘的新方法。提取的规则指示属性值的顺序关系和不同属性的真值的结构。我们还提出了特定的算法来提取广义的单侧量化关联规则和非冗余的广义的单侧量化关联规则。认知计算也需要考虑数据规模。根据不同的测量比例，一个对象可能具有相同属性的不同值。还讨论了不同规模的广义单侧量化关联规则之间的关系。不是将多值形式上下文转换为二进制形式上下文，而是直接在多值形式上下文中提取广义的单面定量关联规则。实验结果表明，与经典的定量关联规则挖掘算法相比，该算法降低了时间和空间复杂度。在多值形式上下文中直接提取广义的单面定量关联规则。实验结果表明，与经典的定量关联规则挖掘算法相比，该算法降低了时间和空间复杂度。在多值形式上下文中直接提取广义的单面定量关联规则。实验结果表明，与经典的定量关联规则挖掘算法相比，该算法降低了时间和空间复杂度。