Named entity recognition with multiple segment representations

作者：

Highlights：

• Different segmentation representations (SRs) cause little difference in performance.

• Different SRs result in quite different outputs.

• Incorporation of different SRs is beneficial to NER task.

• We proposed a new feature generation method that uses multiple SRs.

• The proposed method improves the performance and the stability of NER.

摘要

•Different segmentation representations (SRs) cause little difference in performance.•Different SRs result in quite different outputs.•Incorporation of different SRs is beneficial to NER task.•We proposed a new feature generation method that uses multiple SRs.•The proposed method improves the performance and the stability of NER.

论文关键词：Named entity recognition,Machine learning,Conditional random fields,Feature engineering

论文评审过程：Received 9 April 2012, Revised 3 March 2013, Accepted 5 March 2013, Available online 9 April 2013.

论文官网地址：https://doi.org/10.1016/j.ipm.2013.03.002