摘要： 提出了基于WSTB(Weighted Shape To Bit-vector)的相似搜索方法，该方法在线性分段的基础上建立时间序列曲线箱，而且创立具有相似形状的时序子序列箱后建立相应的索引，对给定序列和相似序列距离的快速计算，并根据查询的时间序列的特征确定相应的权重，不需要逐个检查子序列箱内容就可以进行快速索引。WSTB 方法避免了进行逐个距离比较而造成的巨大的计算量，从而明显地提高搜索效率。最后验证了方法的通用性和有效性。

关键词: 数据挖掘；时间序列；线性分段；相似性

Abstract: A WSTB-based algorithm for similarity search is proposed which is based on the piecewise linear representation. The subsequence bin for time series is built at first and the index of the bin is built. After that, the distance of the given sequence and similar sequence is calculated. The weighted coefficient for every sequence is decided on the character. So the inquiry can be implemented without checking the content of the bin. The quantity of the WSTB calculation which is got from comparing one by one is avoided. The searching efficiency can be improved obviously. At last, the currency and efficiency of the algorithm are proved.

Key words: Data mining; Time series; Piecewise linear representation; Similarity