Unsupervised Boundary-Aware Language Model Pretraining for Chinese Sequence Labeling

27 Oct 2022  ·  Peijie Jiang, Dingkun Long, Yanzhao Zhang, Pengjun Xie, Meishan Zhang, Min Zhang ·

Boundary information is critical for various Chinese language processing tasks, such as word segmentation, part-of-speech tagging, and named entity recognition. Previous studies usually resorted to the use of a high-quality external lexicon, where lexicon items can offer explicit boundary information. However, to ensure the quality of the lexicon, great human effort is always necessary, which has been generally ignored. In this work, we suggest unsupervised statistical boundary information instead, and propose an architecture to encode the information directly into pre-trained language models, resulting in Boundary-Aware BERT (BABERT). We apply BABERT for feature induction of Chinese sequence labeling tasks. Experimental results on ten benchmarks of Chinese sequence labeling demonstrate that BABERT can provide consistent improvements on all datasets. In addition, our method can complement previous supervised lexicon exploration, where further improvements can be achieved when integrated with external lexicon information.

PDF Abstract

Results from the Paper

Task Dataset Model Metric Name Metric Value Global Rank Uses Extra
Training Data
Chinese Word Segmentation CTB6 BABERT-LE F1 97.56 # 1
Chinese Word Segmentation CTB6 BABERT F1 97.45 # 2
Chinese Word Segmentation MSR BABERT-LE F1 98.63 # 1
Chinese Word Segmentation MSR BABERT F1 98.44 # 2
Chinese Word Segmentation MSRA BABERT-LE F1 98.63 # 1
Chinese Word Segmentation MSRA BABERT F1 98.44 # 2
Chinese Word Segmentation PKU BABERT-LE F1 96.84 # 1
Chinese Word Segmentation PKU BABERT F1 96.70 # 2