Emotion Classification in Short English Texts using Deep Learning Techniques

arXiv:2402.16034v1 Announce Type: new
Abstract: Detecting emotions in limited text datasets from under-resourced languages presents a formidable obstacle, demanding specialized frameworks and computational strategies. This study conducts a thorough examination of deep learning techniques for discerning emotions in short English texts. Deep learning approaches employ transfer learning and word embedding, notably BERT, to attain superior accuracy. To evaluate these methods, we introduce the "SmallEnglishEmotions" dataset, comprising 6372 varied short Persian texts annotated with five primary emotion categories. Our experiments reveal …

abstract arxiv classification computational cs.ai cs.cl datasets deep learning deep learning techniques embedding emotion emotions english frameworks languages strategies study text transfer transfer learning type word word embedding