Human-interaction-based Emotion Forecasting (Hi-EF)
收藏资源简介:
Hi-EF数据集由复旦大学开发,专注于基于人际交互的情感预测任务。该数据集包含3069个多层次情境交互样本,每个样本包含丰富的情感相关标签和三种模态信息。数据集的创建过程包括候选样本生成、异常检测和多方面辅助的可靠标注。Hi-EF数据集主要应用于情感计算领域,旨在通过分析交互情境中的情感动态,预测未来情感状态,为情感管理和支持提供新的工具和方法。
The Hi-EF dataset, developed by Fudan University, focuses on the task of interpersonal interaction-based emotion prediction. It comprises 3069 multi-level situational interaction samples, each carrying abundant emotion-related labels and three modalities of information. The construction process of the Hi-EF dataset includes candidate sample generation, anomaly detection, and multi-faceted auxiliary reliable annotation. The Hi-EF dataset is mainly applied in the field of affective computing, aiming to predict future emotional states by analyzing emotional dynamics in interactive scenarios, thereby providing novel tools and methods for emotion management and support.
Hi-EF: Benchmarking Emotion Forecasting in Human-interaction
数据集概述
- 名称: Hi-EF
- 主题: 人类交互中的情感预测基准
数据集内容
- 包含数据集、样本展示和方法流程图
数据集访问
- 数据集已开放访问,链接为:https://drive.google.com/drive/folders/1VvyU4SCmB7ATpcgbLGPVRTtLwweIFYDk?usp=drive_link

- 1Hi-EF: Benchmarking Emotion Forecasting in Human-interaction复旦大学 · 2024年



