CRIS-Yang/Persona-E2-Dataset

Persona-E²（人格-事件到情感）是一个大规模、基于人类标注的数据集，旨在捕捉具有不同人格特质的个体对多样化文本事件的情感反应。该数据集标志着从“文本表达了什么？”到“文本如何让人们感觉？”的关键转变。数据集包含111,996个高质量标注，每个事件由36名标注者独立标注，标注者具有全面的性格特征（MBTI和BFI）。数据集覆盖新闻、社交媒体和生活经验三个主要领域，包含3,111个经过筛选的事件。数据集结构包括三个核心文件：所有标注者的原始标注、基于性格群体的共识标注以及标注者个人资料。数据集收集和标注过程经过严格筛选和控制，包括内容安全过滤、多维LLM评分和专家验证。数据集使用CC BY-NC-SA 4.0许可，仅限于学术研究使用。

Persona-E² (Persona-Event2Emotion) is a large-scale, human-grounded dataset capturing how individuals with measured personality traits emotionally react to diverse text-based events. It marks a crucial shift from what does the text express? to how does the text make people feel? The dataset provides 111,996 high-quality annotations, with each event independently blind-labeled by 36 annotators profiled using both the Myers-Briggs Type Indicator (MBTI) and the Big Five Inventory (BFI). It spans three primary domains (News, Social Media, and Life Experience) containing 3,111 filtered events. The dataset consists of three core files: raw annotations from all annotators, group consensus annotations based on personality clusters, and annotator profiles. The collection and annotation process involved rigorous 3-stage event filtering, controlled human annotation, and ethics/privacy compliance. The dataset is released under the CC BY-NC-SA 4.0 license for academic use only.