Twitter Sentiment Analysis Using LSTM-Dense-Dropout Hybrid Deep Learning Approach

In today’s world, people have been preferably communicating with one another through web-based social media applications like Facebook, Twitter, WhatsApp, and so on. This trend of online communication further upsurged after COVID pandemic at an exponential rate. From the online media applications, we get web-based media information and can actually look at what sentences are contempt and non-disdain utilizing sentiment analysis. With the increasing rate at which huge data is created by Internet users on various platforms, it becomes necessary to analyse and know the sentiments of people. This helps various organizations to take control and do necessary actions accordingly. When something crucial is happening, it helps in taking decisions without hurting the sentiments of the public. In this research, authors used deep learning layers (LSTM layer, dense layer, and dropout layer) to analyse sentiments of people on Twitter, a leading social media platform of sharing views. Authors demonstrated the improved accuracy, i.e. 95% by implementing the proposed hybrid deep learning model (LSTM-dense-dropout) trained with CNN.