A Computational Approach of Recognizing Emotion from Bengali Texts
Hasan Abid Ruposh, Mohammed Moshiul Hoque · 2019
Emotion recognition is the task of determining distinct emotion exhibited in text. In recent year, due to the availability of enormous amount of textual data, specially dogmatic and self expressiveness of text played a significant role to lead focus in this area. This paper presents an emotion recognition technique that can identify six basic emotions from Bengali texts such as happy, sad, anger, fear, surprise, and disgust respectively. We develop a corpus consisting of 1200 emotive words that are used to train the SVM classifier for identifying different emotions. Experimental result shows that the proposed system can recognize emotions with 73% accuracy which is higher than the Naive based approach (60%).