Using Hybrid CNN-LSTM Model for Sentiment Analysis of COVID-19 Tweets

The COVID-19 pandemic has affected many aspects of everyone’s lives, causing problems in health and the economy. People often share their sentiments about the pandemic on social media, particularly on platforms like Twitter. Sentiment analysis involves using various pre-processed tweets and natural language processing (NLP) techniques to classify them as positive, neutral, or negative. It is a demanding task. Platforms with microblogging features have played a crucial role in disseminating information, such as news and emergencies. These data types are rich in sentiments that contribute to the necessary awareness of safety and preparedness. Previous studies have demonstrated that Twitter usage during disasters increases situational awareness and improves disaster response through the tweets of affected users. However, sometimes a single model is insufficient to achieve desired results, necessitating the creation of a hybrid model. This study aimed to develop a sentiment analysis model using a hybrid CNN-LSTM framework to analyze sentiments in tweets during the COVID-19 pandemic. The model utilized a hybrid CNN-LSTM architecture that focused on predicting whether the sentiment was positive or negative. The developed hybrid CNN-LSTM model achieved a testing accuracy of 84.72% and outperformed standalone CNN and LSTM models.