NLG Texts and Judgements

该数据集包含了由不同方法生成的文本（包括人工撰写、基于模板以及基于机器学习的文本），并附有人类对这些文本质量和自然度的评价。此外，数据集还涵盖了在生成文本中检测到的各种错误类别，以及针对质量和自然度的人工评分。规模上，数据集包含了68个模板生成的文本、69个机器学习生成的文本以及68个人工撰写的文本，总计210篇文本及其评价。该数据集的任务是对自然语言生成进行评估。This dataset contains texts generated by different methods, including human-written, template-based, and machine learning-generated texts, accompanied by human evaluations of the quality and naturalness of these texts. Additionally, the dataset covers various error categories detected in the generated texts, as well as human-rated scores for both quality and naturalness. In terms of scale, the dataset includes 68 template-generated texts, 69 machine learning-generated texts, and 68 human-written texts, totaling 210 texts and their corresponding evaluations. The core task of this dataset is to evaluate natural language generation.