Rethinking Masked Language Modeling for Chinese Spelling Correction

TL;DR

This paper improves Chinese Spelling Correction by analyzing model overfitting issues, introducing a diverse benchmark, and proposing a simple masking strategy that enhances language modeling and achieves state-of-the-art results.

Contribution

It identifies overfitting in BERT-based CSC models, introduces the LEMON benchmark for better evaluation, and proposes a simple masking technique to improve model generalization.

Findings

Random masking of 20% non-error tokens improves language modeling.

The proposed method achieves state-of-the-art results on multiple datasets.

The approach enhances out-of-distribution error pattern generalization.

Abstract

In this paper, we study Chinese Spelling Correction (CSC) as a joint decision made by two separate models: a language model and an error model. Through empirical analysis, we find that fine-tuning BERT tends to over-fit the error model while under-fit the language model, resulting in poor generalization to out-of-distribution error patterns. Given that BERT is the backbone of most CSC models, this phenomenon has a significant negative impact. To address this issue, we are releasing a multi-domain benchmark LEMON, with higher quality and diversity than existing benchmarks, to allow a comprehensive assessment of the open domain generalization of CSC models. Then, we demonstrate that a very simple strategy, randomly masking 20\% non-error tokens from the input sequence during fine-tuning is sufficient for learning a much better language model without sacrificing the error model. This technique can be applied to any model architecture and achieves new state-of-the-art results on SIGHAN, ECSpell, and LEMON.