Happy Dance, Slow Clap: Using Reaction GIFs to Predict Induced Affect on Twitter

TL;DR

This paper introduces ReactionGIF, a novel dataset of 30,000 tweets with induced emotion labels derived from reaction GIFs, enabling improved emotion detection and sentiment analysis in NLP.

Contribution

The paper presents an automated method to collect emotion-labeled texts using reaction GIFs and releases the first dataset of its kind for affective computing research.

Findings

Created ReactionGIF, a 30K tweet dataset with induced emotion labels

Established baselines for sentiment and emotion classification tasks

Demonstrated the utility of reaction GIFs for emotion annotation

Abstract

Datasets with induced emotion labels are scarce but of utmost importance for many NLP tasks. We present a new, automated method for collecting texts along with their induced reaction labels. The method exploits the online use of reaction GIFs, which capture complex affective states. We show how to augment the data with induced emotion and induced sentiment labels. We use our method to create and publish ReactionGIF, a first-of-its-kind affective dataset of 30K tweets. We provide baselines for three new tasks, including induced sentiment prediction and multilabel classification of induced emotions. Our method and dataset open new research opportunities in emotion detection and affective computing.