基于互关联后继树的时间序列相似性查询  

A Fast Similarity Query Method Based on Inter-Relevant Successive Trees Model in Time Series

作　　者：曾海泉[1] 宋扬[1] 申展[1] 胡运发[1] 

机构地区：[1]复旦大学计算机与信息技术系数据库中心,上海200433

出　　处：《计算机研究与发展》2004年第2期325-332,共8页Journal of Computer Research and Development

基　　金：国家自然科学基金项目 ( 60 173 0 2 7)

摘　　要：时间序列的相似性查询是分析时间序列变化规律的一种重要方法 ,对于时间序列的分类、预测以及知识发现都具有重要的现实意义 提出了一种基于分段技术的、同时支持可变长度的快速相似性查询方法 其主要思想是 :首先依据序列变化的重要点将序列逐步分段 ,抽取各子段的变化特征 ,通过分类方法将其转变成符号序列 在此基础上 ,引入一种称为互关联后继树的全文索引技术 ,从而实现序列的快速相似性查询 ,其时间复杂度降到了O(L) ,此外 。Time series are an important type of data. Similarity querying in time series is a basic task to analyze the changing trend of time series. In this paper, a novel method is proposed, which supports fast search similar pattern in time series. It first segments time series based on a series of perceptually important points, and then time series are converted into meaningful symbol sequences in terms of the segment's features and MATH categorization. After that, a new index model is designed, which is called inter-relevant successive trees(IRST), to achieve fast similarity retrieval in multiple time series. Compared with the previous methods, the method is more efficient and allows different lengths matching.

关 键 词：时间序列 相似性查询 重要点分段 互关联后继树 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程]