Towards Effective COVID-19 Sentiment Analysis Using Bio-Inspired Feature Optimization

Abstract

Analyzing sentiments in social media content related to COVID-19 presents major challenges, especially given the sheer volume of data and the complexity of text features. These factors often reduce the effectiveness of classification models, making it crucial to apply smart feature selection to boost both accuracy and efficiency. This paper develops a feature selection framework that strategically integrates statistical filtering methods with evolutionary computation techniques, specifically incorporating Genetic Algorithms (GA) and Particle Swarm Optimization (PSO) for optimal feature space reduction—to address these challenges. By focusing on selecting the most meaningful features, our method reduces unnecessary complexity while retaining the information that matters most. Tests carried out on a dataset of COVID-19 tweets show that this approach improves classification accuracy by around 7% compared to standard feature selection methods. These results highlight how combining statistical filtering with bio-inspired optimization can play an important role in improving sentiment analysis, especially during critical situations like the COVID-19 pandemic.

Keywords