基于滑动窗口的流数据聚类算法研究  

Streaming data cluster algorithm based on sliding window

作　　者：胡彧[1] 闫巧梅[2] 

机构地区：[1]太原理工大学测控技术研究所,山西太原030024 [2]太原理工大学计算机与软件学院,山西太原030024

出　　处：《计算机工程与设计》2008年第21期5621-5623,共3页Computer Engineering and Design

摘　　要：为提高对进化数据流的聚类质量及效率,改进了基于滑动窗口的数据流聚类算法,采用聚类特征指数直方图来支持数据处理,减少了直方图结构的维护数,并在复杂度、聚类效果上得到了进一步改善。理论及验证表明,与传统基于界标模型的聚类算法相比,优化算法可获得较好的工作效率、较小的内存开销和快速的数据处理能力,拓展了数据流挖掘技术的应用领域。Streaming data clustering algorithm based on sliding window model with the purpose of high cluster quality and efficiency is improved. In addition, cluster feature histograms are adopted as the data supporting structure. Comparing with traditional cluster algo- rithms, the complexity and the clustering effect is further improved. In addition, the algorithm will be better efficiency, less memory overhead and fast data processing capabilities and expand the application of data streaming mining technologies.

关 键 词：数据流 滑动窗口 聚类 直方图 数据挖掘 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]