Emotion Detection and Recognition from images can be determined by computer vision and deep learning algorithms and is being used in many industry but emotion detection in texts is a new field where we can detect the sentiment of the people by processing public opinion in various platforms like Twitter, Facebook and more. There are 6 emotions in humans according to the facial expression namely happy, sad, fear, disgust, surprise, and anger. This paper explains how we can use various machine learning and deep learning algorithms to detect emotions from the texts. We will be working on 1D-CNN, LSTM and BERT transformer, discussing the algorithms, comparing their performances and discussing the future scope of emotion detection in text on stock analysis and prediction.
To train these models we need some datasets which contain some sentences and labelled emotions. We are using three datasets to train our models which are Daily-Dialogue, Emotion-Stimulus, ISEAR.
Models: 1DCNN, LSTM, BERT