Sentiment Analysis on Multilingual Tweets using XGBoost Classifiers

Twitter is a free social networking microblogging service that allows registered users to send short messages or post so-called tweets. Twitter members can tweet and follow other users' tweets. These tweets are persistent, searchable, and public. Anyone can see tweets on Twitter regardless of whether they are a follower or not. The tweets posted are either good tweets or bad tweets. Sentiment analysis uses natural language processing and computational linguistics techniques to automate the classification of sentiment generated from tweets. In this work, multilingual tweets are collected and these tweets are divided into positive and negative tweets. First, preprocess the tweets to remove unnecessary content from the tweet. Then extract the adjectives that make up the feature vector and then perform sentiment analysis by classifying the tweets using algorithms such as logistic regression. The polarity scores for the multilingual tweets are also calculated. The XGBoost classifier gives better results for the Twitter dataset than the logistic regression model.