MuCGEC: a Multi-Reference Multi-Source Evaluation Dataset for Chinese Grammatical Error Correction
This provides a new dataset for evaluating Chinese grammatical error correction systems, which is incremental but addresses a specific need in natural language processing for Chinese language learners.
The authors introduced MuCGEC, a multi-reference multi-source evaluation dataset for Chinese Grammatical Error Correction containing 7,063 sentences with 2.3 references per sentence, and tested two enhanced models that achieved competitive benchmark performance.
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}.