Analyzing Social Media Sentiment: Twitter as a Case Study

Abstract

This study examines the problem of Twitter sentimental analysis, which categorizes Tweets as positive or negative. Many applications require analyzing public mood, including organizations attempting to determine the market response to their products, political election forecasting, and macroeconomic phenomena such as stock exchange forecasting. Twitter is a social networking microblogging and digital platform that allows users to update their status in a maximum of 140 characters. It is a rapidly expanding platform with over 200 million registered users, 100 million active users, and half of the people log on every day, tweeting out over 250 million tweets. Public opinion analysis is critical for applications, including firms looking to understand market responses to their products, predict political choices, and forecast socio-economic phenomena like bonds. Through the deep learning methodologies, a recurrent neural network with convolutional neural network models was constructed to do Twitter sentiment analysis to predict if a tweet is positive or negative using a dataset of tweets. The applied methods were trained using a publicly available dataset of 1,600,000 tweets. Several model architectures were trained, with the best one achieving a (93.91%) success rate in recognizing the tweets' matching sentiment. The model's high success rate makes it a valuable advisor and a technique that might be improved to enable an integrated sentiment analyzer system that can work in real-world situations for political marketing.