Lexicon information and pre-trained models, such as BERT, have been combined
to explore Chinese sequence labelling tasks due to their respective strengths.
However, existing methods solely fuse lexicon features via a shallow and random
initialized sequence layer and do not integrate them into the bottom layers of
BERT. In this paper, we propose Lexicon Enhanced BERT (LEBERT) for Chinese
sequence labelling, which integrates external lexicon knowledge into BERT
layers directly by a Lexicon Adapter layer. Compared with the existing methods,
our model facilitates deep lexicon knowledge fusion at the lower layers of
BERT. Experiments on ten Chinese datasets of three tasks including Named Entity
Recognition, Word Segmentation, and Part-of-Speech tagging, show that LEBERT
achieves the state-of-the-art results.

本文提出了一种增强 BERT 的方法，称为 LEBERT，通过引入一个词典适配器层，将外部词典知识直接集成到 BERT 的底层中，以便实现深层次的词典知识融合，从而在多项中文序列标注任务中取得了最佳结果。

使用 BERT Adapter 进行词典增强的中文序列标注

Lexicon Enhanced Chinese Sequence Labeling Using BERT Adapter

Although character-based models using lexicon have achieved promising results
for Chinese named entity recognition (NER) task, some lexical words would
introduce erroneous information due to wrongly matched words. Existing
researches proposed many strategies to integrate lexicon knowledge. However,
they performed with simple first-order lexicon knowledge, which provided
insufficient word information and still faced the challenge of matched word
boundary conflicts; or explored the lexicon knowledge with graph where
higher-order information introducing negative words may disturb the
identification. To alleviate the above limitations, we present new insight into
second-order lexicon knowledge (SLK) of each character in the sentence to
provide more lexical word information including semantic and word boundary
features. Based on these, we propose a SLK-based model with a novel strategy to
integrate the above lexicon knowledge. The proposed model can exploit more
discernible lexical words information with the help of global context.
Experimental results on three public datasets demonstrate the validity of SLK.
The proposed model achieves more excellent performance than the
state-of-the-art comparison methods.

该研究提出了基于字符的第二阶 lexicon 知识的模型，该模型能够更好地利用全局上下文来提取词汇信息，从而提高了中文命名实体识别的效果。