Word sense induction using leader-follower clustering of automatically generated lexical substitutes

作者：

Highlights：

• Making lexical substitutes (LS) can improve the performance of WSI systems.

• Combining the vector space and graph clustering outperforms the performance of WSI.

• Post processing make the results much closer to the reality after processing steps.

• The proposed model can be used for completely unseen words.

摘要

•Making lexical substitutes (LS) can improve the performance of WSI systems.•Combining the vector space and graph clustering outperforms the performance of WSI.•Post processing make the results much closer to the reality after processing steps.•The proposed model can be used for completely unseen words.

论文关键词：Word sense induction,Natural language processing,Graph clustering,Clustering refinement,Lexical substitution

论文评审过程：Received 13 January 2020, Revised 17 March 2021, Accepted 3 May 2021, Available online 8 May 2021, Version of Record 20 May 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.115162