Exploring ensemble oversampling method for imbalanced keyword extraction learning in policy text based on three-way decisions and SMOTE

作者：

Highlights：

• Based on SMOTE, we propose a new oversampling method by utilizing 3WD.

• We introduce classification confidence with 3WD to handle the unbalanced data.

• Our proposed oversampling method can be applied to achieve keyword extraction.

• We find supervised methods can achieve better performance in keyword extraction.

摘要

•Based on SMOTE, we propose a new oversampling method by utilizing 3WD.•We introduce classification confidence with 3WD to handle the unbalanced data.•Our proposed oversampling method can be applied to achieve keyword extraction.•We find supervised methods can achieve better performance in keyword extraction.

论文关键词：Three-way decisions,Unbalanced data,Keyword extraction,SMOTE

论文评审过程：Received 24 March 2021, Revised 21 September 2021, Accepted 5 October 2021, Available online 16 October 2021, Version of Record 23 October 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.116051