E-ReDial
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/Superbooming/E-ReDial
下载链接
链接失效反馈官方服务：
资源简介：
该数据集名为E-ReDial，包含了756个对话，每个对话都附有高质量的重写解释，旨在提高对话推荐系统中的可解释性。此外，该数据集还包括了2,058个高质量的解释，每个解释的平均词数为79.2个，并且按照8:2的比例划分为训练集和测试集。在规模上，数据集包含了756个对话，共计12,003个话语。该数据集的任务是生成可解释的推荐。
This dataset is named E-ReDial, comprising 756 dialogues paired with high-quality rewrite explanations, with the goal of enhancing the interpretability of conversational recommendation systems. Furthermore, the dataset includes 2,058 high-quality explanations, each with an average word count of 79.2. It is split into training and test sets at an 8:2 ratio. In terms of scale, the dataset contains 756 dialogues, totaling 12,003 utterances. The core task of this dataset is to generate explainable recommendations.
提供机构：
Superbooming
