Abstract: In order to achieve fast clustering Web service request data and improve the accuracy of clustering, a Web data clustering algorithm with incremental sorting and retrieval interaction is proposed. The algorithm makes Web data clustering be defined in the time sequence, and uses time series incremental clustering method. First, it reduces the complexity of Web data through data in compressed form, then based on service time similarity, time series data clustering is done. Finally, for the problem of the best service task scheduling Web cluster services, through the implementation capacity of the server as a standard the service tasks are assigned. Simulation results show that compared with high-dimensional data grid-based hierarchical clustering algorithm and incremental learning based multi-objective fuzzy clustering algorithm, the algorithm proposed in this paper at the time of clustering, the clustering accuracy, the success rate of all service execution get better results.

Key words: Web data clustering, time series, task scheduling

摘要： 为了实现Web服务请求数据的快速聚类，并提高聚类的准确率，提出一种基于增量式时间序列和任务调度的Web数据聚类算法，该算法进行了Web数据在时间序列上的聚类定义，并采用增量式时间序列聚类方法，通过数据压缩的形式降低Web数据的复杂性，进行基于服务时间相似性的时间序列数据聚类。针对Web集群服务的最佳服务任务调度问题，通过以服务器执行能力为标准来分配服务任务。实验仿真结果表明，相比基于网格的高维数据层次聚类算法和基于增量学习的多目标模糊聚类算法，提出的算法在聚类时间、聚类精度、服务执行成功率上均获得了更好的效果。

关键词: Web数据聚类, 时间序列, 任务调度