基于二进制表示的事务属性挖掘方法

Mining Approach for Transaction Attribute Based on Binary Expression

在关联规则数据挖掘中采用二进制系统易于产生冗余模式。该文提出一种基于二进制事务属性层次划分的两级数据挖掘方法,即MLADM算法。该算法通过高层次模式获取最大可能频繁模式集,在低层次模式中对其进行验证,优先获得长频繁模式。实验结果表明,该算法可以在密集数据集中有效挖掘长模式并避免冗余模式。

Using the method of binary system in association rule data mining may produce some redundancy patterns. This paper proposes an algorithm based on two levels data mining of binary transaction attributes, namely MLADM algorithm. It gets maximum possible frequent patterns set on high level patterns, and validates these patterns on low level patterns to get first long frequent patterns. The experiment on real datasets proves that MLADM is effective on mining long patterns on dense dataset and can avoid producing redundancy patterns.