EmoDet2: Emotion Detection in English Textual Dialogue using BERT and BiLSTM Models
Hani Al-Omari, Malak Abdullah, Samira Shaikh
Abstract
Emotion detection is one of the most challenging problems in the automated understand of language. Understanding human emotions using text without facial expression is considered a complicated task. Therefore, building a machine that understands the context of the sentences and differentiates between emotions has motivated the machine learning community recently. We propose a system to detect emotions using deep learning approaches. The main input to the system is a combination of GloVe word embeddings, BERT Embeddings and a set of psycholinguistic features (e.g. from AffectiveTweets Weka-package). The proposed system (EmoDet2) is combining a fully connected neural network architecture and BiLSTM neural network to obtain performance results that show substantial improvements (F1-Score 0.748) over the baseline model provided by Semeval-2019 / Task-3 organizers (F1-score 0.58).