Discovering patterns of online popularity from time series

作者：

Highlights：

• We propose a novel shape-based clustering algorithm for multidimensional time series.

• We validate the accuracy on the datasets generated from benchmark time series models.

• Our algorithm successfully discloses popularity dynamics for GitHub and Twitter data.

• We make the Python and Matlab implementations of our algorithms publicly available.

摘要

•We propose a novel shape-based clustering algorithm for multidimensional time series.•We validate the accuracy on the datasets generated from benchmark time series models.•Our algorithm successfully discloses popularity dynamics for GitHub and Twitter data.•We make the Python and Matlab implementations of our algorithms publicly available.

论文关键词：Multidimensional time series,Shape-based clustering,Online popularity,Social media

论文评审过程：Received 30 June 2019, Revised 23 February 2020, Accepted 23 February 2020, Available online 24 February 2020, Version of Record 10 March 2020.

论文官网地址：https://doi.org/10.1016/j.eswa.2020.113337