Intensified Emotion and Sentiment Annotation for Words, Emojis, and Slang in English
Shelley Gupta, Apurva Srivastava, Anisha Verma · Procedia Computer Science · 2025
In today’s digital world everything is shifting online. It would have taken a few more years for us to reach here but because of corona the pace of moving on the online mode was accelerated. Everything was an online class, shopping, wedding etc. But the psychology of any human is in the way that he or she judges the other person’s way of speaking and its human nature. The proposed approach has introduced the annotation of more enhanced dictionary consisting of words, emojis, slangs and slang number with 22 intensified emotions. The dictionary has also been annotated with positive, negative and neutral sentiments, so that the machine can help humans in understanding the emotion and sentiment of other people by use of algorithm. The annotated dataset has been trained using 5 machine learning approaches of GRU, CNN, LSTM, Random Forest, SVM coupled with ensemble techniques of bagging, boosting, and stacking, to obtain the optimal results. The ensemble bagging technique and boosting XG Boost algorithm shows an accuracy of 86.08% and 98.09% respectively.