Sentiment Analysis using Machine Learning and Deep Learning
Yogesh Chandra, Antoreep Jana
Abstract
With the increasing rate at which data is created by internet users on various platforms, it becomes necessary to analyze and make use of the data by the Defense and other Government Organizations and know the sentiment of the people. This shall help the organizations take control of their actions and decide the steps to be taken shortly. Added to it, when something crucial is happening in the nation, it is of paramount importance to decide every step without hurting/violating the sentiments of the people. In the era of Microblogging, which has become quite a popular tool of communication, millions of users share their views and opinions on various day-to-day life issues concerning them directly or indirectly through social media platforms like Twitter, Reddit, Tumblr, Facebook. Data from these sites can be efficiently used for marketing or social studies. In this paper, we have taken into account various methods to perform sentiment analysis. Sentiment Analysis has been performed by using Machine Learning Classifiers. Polarity-based sentiment analysis, and Deep Learning Models are used to classify user's tweets as having ‘positive’ or ‘negative’ sentiment. The idea behind taking in various model architectures was to account for the variance in the opinions and thoughts existing on such social media platforms. These classification models can further be implemented to classify live tweets on twitter on any topic.