对演变数据进行关联规则挖掘的新方法

Mining Long Cycle Frequent Association Rules Based on Evolving Data

针对已有经常性周期关联规则在演变数据和周期长度方面的局限性，文章提出一种新的方法，从而可以对演变数据进行经常性长周期关联规则的挖掘。这种方法针对演变数据的动态环境，通过对数据块的动态聚类得到周期分段，然后在每一分段内利用低支持度关联规则挖掘算法来发现周期较长的关联规则。整个算法可以在GEMM算法的基础上进行动态模式保持。

To overcome the limitation of existing model on evolving data and long cycle, this paper proposes a new method which can discover long cycle frequent association rules based on evolving data. Under the dynamic circumstance of evolving data ,the cyclic segment can be obtained through dynamic clustering analysis. In each segement, some algorithms for mining association rules with low support are used to discover the long cycle association rules. The whole algorithm can be maintained dynamically on the basis of GEMM algorithm .