PersonalityEvd
收藏资源简介:
PersonalityEvd数据集由中国人民大学信息学院构建,包含72名说话者和约2000个来自中国电视剧的对话。该数据集旨在支持可解释的人格识别任务,通过对话内容揭示人格特质的支持证据。数据集内容包括对话级别的个性状态标签和说话者级别的个性特质标签,以及详细的推理过程。创建过程涉及使用GPT-4进行预标注,然后由心理学专业学生进行手动校正。该数据集主要应用于心理学诊断、人机交互等领域,旨在解决传统人格识别方法缺乏解释性的问题。
The PersonalityEvd dataset was constructed by the School of Information, Renmin University of China. It consists of 72 speakers and approximately 2,000 dialogues sourced from Chinese TV dramas. This dataset is designed to support interpretable personality recognition tasks, which uncover supporting evidence for personality traits through dialogue content. The dataset includes dialogue-level personality state labels, speaker-level personality trait labels, as well as detailed reasoning processes. Its construction involved pre-annotation using GPT-4, followed by manual correction by psychology majors. It is primarily applied in fields such as psychological diagnosis and human-computer interaction, aiming to address the lack of interpretability in traditional personality recognition methods.




