Sentiment Analysis of the Russia-Ukraine War Using Twitter Data based on Hybrid Deep Learning Models

Abstract

The Russia-Ukrainian War denotes an ongoing conflict between Russia and Ukraine. When Russia initiated it in February 2014, the primary focus was on whether or not Crimea and the Donbass were officially regarded as parts of Ukraine. After a military build-up on the Russian–Ukrainian border that began in late 2021, the conflict escalated dramatically when Russia initiated its assault of Ukraine on February 24, 2022. The objective of this piece is to investigate how the general population views the situation between Russia and Ukraine. Social media has become a major communication tool these days, and as a result, opinions can be found on sites like Facebook, Instagram, and Twitter. The study uses 11,250 of his tweets from his Twitter account on the conflict between Russia and Ukraine. Machine learning has demonstrated the applicability, strength, and potential of techniques such as object recognition, natural language processing, and image processing. Three feature extraction techniques, TF-IDF (term frequency-inverse document frequency), BoW (bag of words), and N-gram, have been used in the development and testing of our models. The Twitter API is used to build and test hybrid deep sentiment analysis learning models, which integrate support vector machines (SVM), recurrent neural networks (RNN), convolutional neural networks (CNN), and long short-term memory (LSTM) networks. When compared to single models on the Twitter API dataset, the hybrid models particularly the combination of deep learning models and SVM—improved sentiment analysis accuracy.