MuCGEC (Multi-Reference Multi-Source Evaluation Dataset for Chinese Grammatical Error Correction)

MuCGEC 是一个用于汉语语法纠错 (CGEC) 的多参考多源评估数据集，由从三个不同的汉语作为第二语言 (CSL) 学习器来源收集的 7,063 个句子组成。每个句子都由三位注释者更正，他们的更正由专家仔细审查，每个句子有 2.3 个参考文献。

MuCGEC is a multi-reference and multi-source evaluation dataset for Chinese Grammatical Error Correction (CGEC). It consists of 7,063 sentences collected from three distinct sources of Chinese as a Second Language (CSL) learners. Each sentence was corrected by three annotators, whose corrections were carefully reviewed by experts, resulting in an average of 2.3 reference correction versions per sentence.

搜集汇总

背景与挑战

背景概述

MuCGEC是一个专为汉语语法纠错设计的评估数据集，包含7,063个从多个汉语学习来源收集的句子，每个句子都经过三位注释者更正和专家审查，平均提供2.3个参考更正。该数据集由苏州大学和阿里巴巴于2022年发布，用于支持多参考多源评估。