Opinion Analysis of Bi-Lingual Event Data from Social Networks

Abstract

In the internet age, social media has immediately connected all of us. Social media platforms connect people. Thanks to the social network Twitter, which enables users to tweet their opinions on any specific event or political organisation, we may access a vast range of political information. This study’s objective is to use natural language processing to process a dataset (NLP). The actions are as follows: obtaining information from Twitter and using sentiment analysis, looking into deep learning approaches for training models and sentiment analysis, and supplying a Python library that categorizes input texts as positive or negative, and doing so. Languages were represented in the training data: Roman Urdu (89793). To categorise feelings, various classification models are used, and ultimately, the ensemble technique is investigated in order to go forward with the collected results. The LSTM classifier had an accuracy of 87%, while the Bert model fared best with an accuracy of 90%.