PGDataset

该数据集包含了与相关话语对齐的简介句子，这些句子是从对话语料库中提取出来的，用于生成个人简介任务。此外，通过人工评估自动注释的质量，根据蕴含关系的softmax概率来衡量置信度。该任务的目标是从对话中生成个人简介。

This dataset contains brief profile sentences aligned with relevant utterances, which are extracted from conversational corpora and designed for the personal profile generation task. Additionally, the quality of automatic annotations is evaluated manually, with confidence measured by the softmax probability of the entailment relationship. The goal of this task is to generate personal profiles from dialogues.