Bangla Text Emotion Classification using LR, MNB and MLP with TF-IDF & CountVectorizer

Abstract

Emotions have a significant role in human contact. Written material, vocal discourse, and facial expressions can all be used to convey emotion. The habit of showing emotion on digital platforms or blogs has grown considerably in recent years. Bangla is a widely spoken language throughout the globe, with billions of people speaking it. Bangla is used by these folks to express their emotions. It will be wonderful to have a means to identify these emotions outside of the text. In order to achieve this goal, we tested three algorithms for detecting emotion in Bangla texts. Logistic Regression, Multinomial Naive Bayes, and Multi-layer Perceptron have all been used to determine six identical emotion-related categories. TF-IDF, count vectorizer and their combination is used as features on two blended datasets to evaluate the performance of these three algorithms. It is found that he LR with TF-IDF approach gives the best overall accuracy, precision, recall, and F1-measure score among all of the results.