【作者】 张龙波； 李战怀； 余敏； 王勇； 蒋芸；

【机构】 西北工业大学计算机学院；

【摘要】 <正>1引言近年来在很多应用领域中出现了一种新的数据模式,其数据不是以传统的有限数据集形式,而是连续的数据流形式出现。数据流中的数据基本元素仍然可能是关系元组,但数据的到达是快速、时变、不可预测和无限的数据流形式,不可能完全存储原始更多还原

【Abstract】 Recently,there have been more and more interests in data stream management system and algorithms over data streams.This paper presents two random sampling algorithms for processing data streams-SBWRS algorithm and TBWSS algorithm.SBWRS algorithm,which is a basic window-based sampling algorithm,extends classic reservoir sampling to deal with the expiration of data elements from sequence-based sliding window.TBWSS algorithm, which is a kind of stratified sampling,samples different data items with unequal probabilities from timestamp -based sliding window.The theoretic analysis and experiments show that the algorithms are effective and efficient for continuous data streams processing.更多还原