摘要
传统的类Apriori频繁序列模式挖掘算法都是基于支持度框架理论,需要预先设定支持度阈值,而这通常需要较深的领域知识或大量的实践,因此目前仍没有一种很好的设定方法。同时,序列模式的挖掘结果往往数量很大且不易理解,可用性较低。针对上述问题,提出了一种基于逻辑的频繁序列模式挖掘算法即LFSPM算法,并首次在频繁序列模式挖掘算法中引入了逻辑的思想,通过逻辑规则过滤,大大优化了结果集。实验证明,该算法较好地解决了支持度设置问题及挖掘结果可理解性不高的问题。
Traditional Apriori-like sequential pattern mining algorithms are based on the theoretical framework of support, which need pre-set support threshold, but this often requires in-depth domain knowledge or a lot of practice. Consequently,there is still no good way to set it. Meanwhile, the results of sequential patterns are too large to understand and apply. To solve these problems,this paper presented a logic-based frequent sequential pattern mining algorithm LFSPM, and introduced the thought of logic into frequent pattern mining process for the first time. Through using logical rules to filter, it optimizes the result sets greatly. Experiments show good performance of the proposed approach to solve these problems.
出处
《计算机科学》
CSCD
北大核心
2015年第5期260-264,共5页
Computer Science
基金
浙江省自然科学基金(LY14F020018)
国家自然科学基金(61202204)资助
关键词
频繁序列模式
数据挖掘
逻辑
支持度阈值
Frequent sequential pattern,Data mining,Logic,Support threshold
