基于传感器网络的关联规则挖掘算法研究

现有的传感器网络数据处理系统只能向用户提供简单的查询操作,如何高效地处理传感器网络的海量数据流,从中获取有用的知识成为新的挑战.提出一种基于传感器网络特性的分布式关联规则挖掘算法,采用树型通信结构,各个传感器通过改进的单一数据流频繁项集挖掘算法找出本地的局部频繁项集,并逐层上传、合并,最后由sink节点将所有子节点的局部频繁项集合并成全局的频繁项集,并产生相应的关联规则.实验证明该算法占用较少的计算时间和内存.

Research on association rules mining arithmetic based on sensor networks

Existing data processing system of sensor networks provide with simply query only, how to process the large data streams in sensor networks efficiently and how to find interesting knowledge in these streams become new challenge. An algorithm of mining association rules in distributed data streams of sensor networks are developed. Adopt tree-type communicate structure, each sensor finds local frequent itemset by improving frequent itemset mining arithmetic in single data stream, transmits and combines, at last combineed into a global frequent itemset by central node, association rules are produced based on them. Test prove this arithmetic spend less calculate time and memory.