Cross Domain Emotion Recognition using Few Shot Knowledge Transfer (2110.05021v1)

Abstract: Emotion recognition from text is a challenging task due to diverse emotion taxonomies, lack of reliable labeled data in different domains, and highly subjective annotation standards. Few-shot and zero-shot techniques can generalize across unseen emotions by projecting the documents and emotion labels onto a shared embedding space. In this work, we explore the task of few-shot emotion recognition by transferring the knowledge gained from supervision on the GoEmotions Reddit dataset to the SemEval tweets corpus, using different emotion representation methods. The results show that knowledge transfer using external knowledge bases and fine-tuned encoders perform comparably as supervised baselines, requiring minimal supervision from the task dataset.