缩小候选集的top-k高效模式挖掘算法

Top-k utility Pattern Mining With Decreased Candidate

在通常的模式挖掘中,为了筛选出有效模式,用户需要设置阈值。但是,如何设定一个合适的阈值却是一件困难的事情。Top-k高效模式挖掘算法避免设置阈值,同时考虑了现实数据的一些属性的重要性。尽管相关算法近年已经提出,但是往往会产生大量的候选模式。本文提出了一种挖掘k个最有价值模式的算法,并且不会产生太多的候选项。它通过伺机选择阈值提高策略,从而有效缩小在挖掘过程中的候选集大小。

In the usual pattern mining,in order to find out the utility pattern,the user needs to set a threshold.But to set an appropriate threshold is dif icult.Top-k efficient pattern mining algorithm avoids setting threshold,taking into account the importance of some properties of real data.Although related algorithms have been proposed in recent years,but they tend to produce a large number of candidate patterns.This paper presents a top-k high utility pattern mining algorithms,and does not produce too many candidates.It increasesthreshold by opportune select strategy,which can ef ectively reduce the candidates set during mining process.