Various Twitter Sentiment Analysis Topics using Feature Reduction method based on Binary Gray Wolf Optimizer with S-shape Transfer Function

Sentiment analysis is a field of text mining in which text data about the consumer's feelings or attitudes is collected using various approaches. However, it is common for text data to contain noisy and irrelevant features. Therefore, to enhance the sentiment analysis process, it is necessary to use a method that reduces the number of dimensions in the data and determines the most important features. To address these challenges, Binary Gray Wolf Optimization (BGWO) was utilized in this study for feature selection optimization. The classification technique employs the K-Nearest Neighbors (KNN), Support Vector Machine (SVM), and Naive Bayes (NB) classifiers. Seven distinct Twitter datasets with varying subjects are utilized to validate the suggested approach's performance. The suggested approach is compared to a sentiment feature selection method based on a genetic algorithm. We also examined the performance of the three classifiers before and after using the sentiment feature selection strategy. The findings of comprehensive trials show that the suggested approach surpasses others in terms of accuracy, precision, recall, and F-measure. The findings show that the suggested approach is resilient, as it decreased the number of features by up to 90% while improving accuracy in all datasets.