Abstract: According to the feature that dynamic trend degree of association rules can change over time, this paper puts forward a method of applying the grey-Markov model to tendency measure mining in dynamic association rules on the basis of analyzing the original definition and the dynamic correlation trend prediction model. The trend of the rules can be obtained with the definition of the tendency measure mining. For those rules which do not meet the threshold value of trend, the method uses grey-Markov model to forecast their support degree counting sequences. And then the predicted data is joined into Meta-rule’s support degree counting sequences to get the new trend of the rules which should be determined whether it can meet the threshold requirements. The method is not only proved to be valid but also improving the accuracy and efficiency of mining so that dynamic association rule mining can get more comprehensive and more accurate results through analyzing a case.

Key words: data mining, dynamic association rules, tendency measure, Grey-Markov（GM）（1, 1）model, Markov chain, forecast

摘要： 针对动态关联规则趋势度随时间变化的特点，在分析原有定义以及对动态关联规则趋势度建立预测模型的基础上，提出一种把灰色-Markov模型应用到动态关联规则趋势度挖掘中的方法。该方法利用动态关联规则趋势度定义得到规则的趋势度；对于不满足趋势度阈值的规则的支持度计数序列运用灰色-Markov模型进行预测；将预测数据添加到原规则支持度序列中，并且得到该规则新的趋势度，进而判定此规则的趋势度是否满足阈值要求。通过一个实例进行分析，结果不仅证明了该方法的有效性并且能在一定程度上提高了挖掘的精度和效率，从而使动态关联规则挖掘能够得到更全面、更精确的结果。

关键词: 数据挖掘;动态关联规则;趋势度;灰色-Markov（GM）（1, 1）模型;Markov链;预测