一种面向分布式数据流的闭频繁模式挖掘方法

Closed frequent patterns mining method over distributed data streams

对智能交通系统中面向分布式数据流的频繁模式挖掘问题进行了研究。针对智能交通系统中传感器网络数据流的特点,提出一种基于分布式窗口树的分布式数据流闭频繁模式挖掘方法。该方法在分布式节点中构建分布式窗口树,通过对分布式窗口树进行更新、剪枝及挖掘,能够快速响应用户的查询请求,返回任意时间窗口内数据中的闭频繁模式。实验表明,在保证挖掘准确性的前提下,该方法能够有效缩短查询响应时间,并具有良好的可扩展性。

The paper concentrated on frequent patterns mining problem over distributed data streams in intelligent transportation systems. According to the characteristics of sensor network data streams in intelligent transportation systems, this paper proposed a method for mining closed frequent patterns in arbitrary time window of distributed data streams. As data stream flows, the method captured the contents of data streams with a distributed compact prefix-tree, which was preserved in the distributed computing nodes. It deleted the obsolete and infrequent items by periodically pruning the tree. With mining the tree paralled, the result could be returned to user rapidly after submitting the query. The experimental results show that, ensuring the accuracy of mining, the method can effectively reduce the query response time, and has good scalability.