A FRAMEWORK FOR OPTIMIZING AN ENSEMBLE LEARNING FOR OFFENSIVE TWEET IDENTIFICATION

Abstract

Online derogatory comments are ubiquitous on social media (SM) and are raising serious concerns. SM data is ridden with high dimensional search space due to noise, redundant features, and non-standardized writing style. These problems lead to high computational cost, longer training time with overall low predictive accuracy in machine learning models. To address these problems, we proposed a framework for optimizing stacked generalized ensemble learning to enhance the model performance. The main components of the framework include feature optimizer (FO), ensemble classifiers, and stratified K-foldCV (skfCV) through stacked generalization ensemble architecture. The ensemble classifiers, FO, and skfCV components make our method stable, computationally efficient with the best performance. Our proposed method was validated using three benchmark datasets. The proposed method outperformed the state-of-the-art results in all the evaluation metrics used in those three articles adopted for comparison.