Detecting dynamic community by fusing network embedding and nonnegative matrix factorization

作者：

Highlights：

• We prove the equivalence between network embedding and NMF, providing the theoretical foundation for algorithms.

• We propose a novel NE2NMF algorithm for dynamic community in temporal networks.

• The experimental results demonstrate that NE2NMF not only outperforms state-of-the-art approaches but also reduces the running times of the algorithm.

摘要

•We prove the equivalence between network embedding and NMF, providing the theoretical foundation for algorithms.•We propose a novel NE2NMF algorithm for dynamic community in temporal networks.•The experimental results demonstrate that NE2NMF not only outperforms state-of-the-art approaches but also reduces the running times of the algorithm.

论文关键词：Dynamic community detection,Temporal networks,Nonnegative matrix factorization,Network embedding,Trace optimization

论文评审过程：Received 24 August 2020, Revised 12 March 2021, Accepted 15 March 2021, Available online 20 March 2021, Version of Record 27 March 2021.

论文官网地址：https://doi.org/10.1016/j.knosys.2021.106961