Ranking and significance of variable-length similarity-based time series motifs

作者：

Highlights：

• Motifs of variable-length are becoming an important asset of current expert systems.

• We show that such motifs cannot be compared using length-normalized dissimilarities.

• We propose a solution to rank variable-length motifs and measure their significance.

• It relies on a compact dissimilarity space model based on the beta distribution.

摘要

•Motifs of variable-length are becoming an important asset of current expert systems.•We show that such motifs cannot be compared using length-normalized dissimilarities.•We propose a solution to rank variable-length motifs and measure their significance.•It relies on a compact dissimilarity space model based on the beta distribution.

论文关键词：Time series,Motif ranking,Distance modeling,Beta distribution

论文评审过程：Received 5 March 2015, Revised 11 February 2016, Accepted 12 February 2016, Available online 27 February 2016, Version of Record 11 March 2016.

论文官网地址：https://doi.org/10.1016/j.eswa.2016.02.026