摘要： 数据流是带有时间属性、持续到达的数据序列，实际中有不同的应用形式。采用应用驱动的研究思路，从具体应用中抽象出4种不同的数据流模型，提取出数据维数和子流个数等2个参数，将各种数据流模型在表示形式上统一起来，分析不同数据流模型的聚类方法，探索并给出一个基于滑动窗口的统一数据流模型聚类挖掘框架。

关键词: 数据流, 应用, 聚类, 表示模型, 统一

Abstract: Data stream is data sequence with time attribute, and the data continually arrive. There are different applied forms of data stream in practical applications. This paper uses the research approach of application-driven, abstracts various stream models from the concrete applications, extracts two parameters which unify the different data stream models on representation, analyzes the clustering methods of different data stream models, explores and gives a clustering frame of the unified data stream model based on sliding window.

Key words: data stream, application, clustering, representation model, unified

中图分类号: 

• TN919.2