滑动窗口模型下的优化数据流聚类算法  

Optimal algorithm of data streams clustering on sliding window model

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,太原030024 [2]太原理工大学计算机与软件学院,太原030024

出　　处：《计算机应用》2008年第6期1414-1416,共3页journal of Computer Applications

基　　金：国家自然科学基金资助项目(60374029);山西省教育厅高校科技研究开发项目(101049)

摘　　要：为提高对进化数据流的聚类质量及效率,采用聚类特征指数直方图支持数据处理,减少直方图结构的维护数,改进滑动窗口下的流数据聚类算法。实验表明,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,拓展了流数据挖掘技术的应用领域。Streaming data clustering algorithm was improved based on sliding window model with the purpose of high cluster quality and efficiency. The new algorithm adopted cluster feature histograms as the data supporting structure. Compared with traditional cluster algorithms, the complexity and clustering effect of new algorithm was further improved. In addition, the algorithm is of better efficiency, less memory overhead and fast data processing capabilities, and the application of data streaming mining technologies is expanded.

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]