Vent

该数据集包含了来自Vent社交应用上的超过3300万条帖子，每条帖子都由作者标注了情绪标签。经过预处理，数据集包含了超过900万条帖子，这些帖子被标记为五种情绪状态中的一种：悲伤、愤怒、恐惧、快乐和喜爱。在预处理过程中，非英文帖子、重复内容和无关信息已被移除。情绪标签根据情感科学文献被映射到语言类别。该数据集的规模超过900万条帖子，其任务是文本中的情绪识别。

This dataset includes over 33 million posts sourced from the Vent social application, with every post annotated with emotion labels by its original author. After preprocessing, the dataset retains more than 9 million posts, each assigned to one of five emotional states: sadness, anger, fear, joy, and love. During the preprocessing stage, non-English posts, duplicate content, and irrelevant information were filtered out. Emotion labels are mapped to linguistic categories in accordance with affective science literature. Comprising over 9 million posts, this dataset is developed for the task of text-based emotion recognition.

搜集汇总

背景与挑战

背景概述

Vent数据集是一个用于研究大规模情感分享的文本数据集，发布于2019年，数据体积为330.2 GB。该数据集访问受限，仅限研究和非商业用途，需联系作者申请获取。