Abstract: The types of medicine are dazzling, however, positions in medicine vending machine are limited. In order to buy medicine conveniently, an idea using association rules to warehouse is put forward. Firstly, the essay introduces insufficient on traditional metric methods. Secondly, in order to make sure positive correlation among one item with others, correlation metric is proposed. Thirdly, relevance metric is proposed to delete cross support item-sets. Finally, the essay puts forward an algorithm called SRL using support metric, correlation metric and relevance metric. The experiment indicates SRL algorithm’s time complexity and effectiveness are better than traditional Apriori algorithm’s. In other words, SRL algorithm is feasible.

Key words: medicine vending machine, warehouse, association rule, interest measure

摘要： 药品种类琳琅满目，自动售药机的仓位却是有限的。针对如何配仓才能方便购买药品的问题，提出使用关联规则挖掘频繁项集配仓的想法。首先介绍了传统度量方法的不足；其次考虑项与项之间的正负相关性，引入相关度的概念，保证项与项之间都是正相关的；然后针对存在交叉支持模式的问题，引入关联度的概念，去除交叉支持模式；最后针对实际的药品销售数据集，结合支持度、相关度、关联度提出SRL算法。实验表明SRL算法的时间复杂度和有效性都优于传统的Apriori算法，证明了SRL算法的可行性。

关键词: 自动售药机, 配仓, 关联规则, 兴趣度