AGAIN数据集

AGAIN数据集是由马耳他大学数字游戏研究所创建的大规模情感语料库，包含来自九款不同游戏的超过1,100个游戏视频（及其相应的游戏数据），由124名参与者以第一人称连续方式进行唤醒注释。该数据集旨在研究不同任务间情感计算的普遍性，是目前基于游戏作为交互式情感触发器的最大和最多样化的公开可用情感数据集，总时长超过37小时。数据集不仅适用于研究每款特定互动游戏的情感建模，还可用于研究游戏中的普遍情感检测，拓宽了通用AI表示和人工通用智能研究的视野。The AGAIN Dataset is a large-scale affective corpus created by the Digital Games Institute of the University of Malta. It contains over 1,100 gameplay videos from nine distinct games along with their corresponding game data, which were annotated for arousal in a first-person continuous manner by 124 participants. This dataset aims to investigate the generalizability of affective computing across different tasks, and is currently the largest and most diverse publicly available affective dataset that uses games as interactive emotion triggers, with a total duration of over 37 hours. It is not only applicable to research on affective modeling for each specific interactive game, but also can be used to study generalizable emotion detection in games, broadening the horizons of research on general AI representation and artificial general intelligence (AGI).

搜集汇总

背景与挑战

背景概述

AGAIN数据集是一个由马耳他大学创建的大规模情感语料库，包含来自九款游戏的超过1,100个视频和游戏数据，总时长超37小时，由124名参与者进行唤醒注释。它旨在研究游戏情感计算的普遍性，是目前最大、最多样化的公开游戏情感数据集，适用于特定游戏情感建模和普遍情感检测，支持通用AI和人工通用智能研究。