MuCGEC: a Multi-Reference Multi-Source Evaluation Dataset for Chinese Grammatical Error Correction

TL;DR

MuCGEC introduces a comprehensive Chinese grammatical error correction dataset with multiple references, enabling more accurate evaluation of CGEC models, and demonstrates the effectiveness of large pretrained models on this benchmark.

Contribution

The paper provides a new multi-reference, multi-source dataset for Chinese GEC, along with evaluation insights and benchmark results using advanced models.

Findings

Large pretrained models improve CGEC performance.

Multiple references enhance evaluation accuracy.

Benchmark results set new standards for Chinese GEC.

Abstract

This paper presents MuCGEC, a multi-reference multi-source evaluation dataset for Chinese Grammatical Error Correction (CGEC), consisting of 7,063 sentences collected from three Chinese-as-a-Second-Language (CSL) learner sources. Each sentence is corrected by three annotators, and their corrections are carefully reviewed by a senior annotator, resulting in 2.3 references per sentence. We conduct experiments with two mainstream CGEC models, i.e., the sequence-to-sequence model and the sequence-to-edit model, both enhanced with large pretrained language models, achieving competitive benchmark performance on previous and our datasets. We also discuss CGEC evaluation methodologies, including the effect of multiple references and using a char-based metric. Our annotation guidelines, data, and code are available at \url{https://github.com/HillZhang1999/MuCGEC}.