Sentiment Analysis of General Peoples Reaction About Covid19 Vaccination in Bangladesh Using Machine Learning Algorithm from Bengali Text Dataset

Abstract

Nowadays, Sentiment analysis is among the most advance discussable topics in Natural Language Processing field. Sentiment analysis identifies a paragraph's specific pole. Currently, Covid-19 pandemic is one of the most outrageous disease facing by humans. Bangladesh is also suffering by this disease. After detecting the first case of covid. The Government is attempting to coordinate the delivery, vaccination, and distribution of Covid-19 vaccines. Then when the procedure of vaccination starts, the question arises in the minds of the general public whether the vaccine will be good or not? Here, utilizing various classification analysis algorithms based on machine learning, we aim to extract sentiment from the Bengali paragraph, which is ordinary people's reaction to the covid vaccination process. Before, starting the process we have studied various research paper, journal and other online articles to gather knowledge about the process of sentiment analysis. We gathered data for this work from an online survey, multiple social networking sites, and other sources and classify them by Positive, Negative and Neutral class. Preprocessing Bengali text is one of the most difficult aspects of the entire process. We had to overcome several obstacles in order to achieve a satisfactory result. Here, We have implemented six popular classification algorithm which are Naïve Bayes, Random forest, SVM, Decision Tree, K-nearest neighbors, Logistics Regression and two deep learning algorithm, which are LSTM and CNN. Among them CNN provide us the maximum accuracy which is 65.41%.