EMOTION RECOGNITION BY TEXTUAL TWEETS CLASSIFICATION USING VOTING CLASSIFIER

Abstract

Due to the abundance of user-generated content on social media sites, point of view extraction has become a challenging task. In order to gather views regarding products, trends, and national politics, people use Twitter, a micro blogging network. Applying belief evaluation, a method for gauging how different individuals feel and think about a given topic, to tweets allows one to see how the public feels about particular news stories, legislation, social movements, and personalities. It is now possible to do opinion mining without manually scanning tweets by using variations of artificial intelligence. The policies, products, and events that federal governments and organisations showcase could benefit from their findings. By dividing tweets into happy and sad categories, seven ML models are used for emotion detection.The suggested ballot classifier (LR-SGD) with TF-IDF produced the best results (79% accuracy and 81% F1 score) in a thorough comparative efficiency study. In order to confirm the stability of the suggested approach on two more datasets, one with binary data and the other with multi-class data, and to achieve long-term results.