摘要： 为发现处在低支持度下的潜在有趣模式，针对传统基于支持度策略的模式发现算法存在的问题，提出一种基于改进Relim算法的超团模式挖掘算法，将一个事务拆分为2个或多个事务，把相同事务进行压缩，并用Relim算法的思想进行超团模式挖掘。仿真实验结果表明，该算法能有效提高超团模式的挖掘效率。

关键词: 数据挖掘, 关联规则, 超团模式, 事务拆分, Relim算法

Abstract: In order to discover potential interesting patterns at low levels of support, aiming at problems in traditional support degree-based mode discovering algorithm, this paper proposes a hyperclique mining algorithm based on improved Relim algorithm. It splits a transaction into two or more transactions and compresses the same transactions. By using the idea of Relim algorithm, the hyperclique pattern is mined. Simulation experimental results show this algorithm can promote the mining efficiency for hyperclique pattern.

Key words: data mining, association rules, hyperclique pattern, transaction splitting, Relim algorithm

中图分类号: 

• TP301.6