DyLex: Incorporating Dynamic Lexicons into BERT for Sequence Labeling
This addresses the problem of efficiently integrating frequently updated lexicons into deep learning models for sequence labeling tasks, offering a practical solution for NLP applications.
The paper tackles the challenge of incorporating large-scale dynamic lexicons into BERT for sequence labeling, which often causes matching noise and update issues, and proposes DyLex, a plug-in approach that achieves new state-of-the-art results on ten datasets across three tasks.
Incorporating lexical knowledge into deep learning models has been proved to be very effective for sequence labeling tasks. However, previous works commonly have difficulty dealing with large-scale dynamic lexicons which often cause excessive matching noise and problems of frequent updates. In this paper, we propose DyLex, a plug-in lexicon incorporation approach for BERT based sequence labeling tasks. Instead of leveraging embeddings of words in the lexicon as in conventional methods, we adopt word-agnostic tag embeddings to avoid re-training the representation while updating the lexicon. Moreover, we employ an effective supervised lexical knowledge denoising method to smooth out matching noise. Finally, we introduce a col-wise attention based knowledge fusion mechanism to guarantee the pluggability of the proposed framework. Experiments on ten datasets of three tasks show that the proposed framework achieves new SOTA, even with very large scale lexicons.