ErAConD
收藏资源简介:
ErAConD是由哥伦比亚大学和浙江大学合作开发的数据集,专注于人机对话场景下的语法错误修正。该数据集包含1735条来自开放领域聊天机器人的对话,每条对话都经过详细标注,以反映真实世界语言学习应用的需求。数据集的创建过程涉及使用BlenderBot在Amazon Mechanical Turk上收集对话,并由英语母语者进行手动修正和标注。ErAConD的应用领域主要集中在改进教育聊天机器人的语法错误反馈,特别是在非正式写作和对话场景中。
ErAConD is a dataset co-developed by Columbia University and Zhejiang University, focusing on grammatical error correction in human-machine dialogue scenarios. This dataset includes 1,735 dialogues from open-domain chatbots, with each dialogue thoroughly annotated to meet the requirements of real-world language learning applications. The dataset's development process involved collecting dialogues via BlenderBot on Amazon Mechanical Turk, followed by manual correction and annotation by native English speakers. The main application fields of ErAConD focus on improving grammatical error feedback for educational chatbots, particularly in informal writing and dialogue scenarios.

- 1ErAConD : Error Annotated Conversational Dialog Dataset for Grammatical Error Correction哥伦比亚大学 · 2022年



