An algorithm based on density and compactness for dynamic overlapping clustering

作者：

Highlights：

• A new dynamic overlapping clustering algorithm, called DClustR, is proposed.

• DClustR introduces a new graph-covering strategy and a new filtering strategy.

• DClustR solves limitations of the state-of-the-art algorithms.

• DClustR outperforms state-of-the-art algorithms in several collections.

• DClustR has a better quality-efficiency trade off than state-of-the-art algorithms.

摘要

Highlights•A new dynamic overlapping clustering algorithm, called DClustR, is proposed.•DClustR introduces a new graph-covering strategy and a new filtering strategy.•DClustR solves limitations of the state-of-the-art algorithms.•DClustR outperforms state-of-the-art algorithms in several collections.•DClustR has a better quality-efficiency trade off than state-of-the-art algorithms.

论文关键词：Data mining,Clustering,Overlapping clustering algorithms,Dynamic clustering algorithms

论文评审过程：Received 5 June 2012, Revised 25 March 2013, Accepted 27 March 2013, Available online 10 April 2013.

论文官网地址：https://doi.org/10.1016/j.patcog.2013.03.022