Particle Swarm Optimization (PSO) approach for feature selection in sentiment analysis for propaganda issues

Feature selection, one of the main components of feature engineering, is the process of selecting the most important features to be input in machine learning algorithms. Feature selection techniques are employed to diminish the number of input variables by eliminating redundant or irrelevant features and narrowing down the set of features without deducting the predictive efficiency. The polynomial of feature selection in sentiment analysis is the main problem because it can decelerate sentiment classification accuracy and misfortunes to extort the optimum of features’ subset. To overcome the problem, this research provides a text feature selection approach that requires an optimization method. Particle swarm optimization (PSO), one of the optimization approaches has been applied as feature selection in this research. PSO is the bio-inspired algorithm, and it is a simple classifier to search for an optimal solution in the solution space. It is different from other optimization algorithms in such a way that only the objective function is needed, and it is not dependent on the gradient, domain, or any differential form of the objective. It also has very few hyperparameters. This research presents the implementation of the PSO (feature selection) performance in the propaganda domain for sentiment analysis. The effectiveness of the PSO algorithm is tested using the datasets from Kaggle concerning Donald Trump and Hillary Clinton's tweets during US Election Presidential in 2016. Three experimental results have shown that PSO is better and more substantial in constructing optimum and quality feature subsets compared to other sentiment analysis tools, machine learning algorithms and swarm algorithms. Suggestion of the PSO algorithm implementation in others domain as tourism, medicine, product reviews etc.