Determining an Optimal Data Classification Model for Credibility-Based Fake News Detection

Enhancing the Predictive Performance of Credibility-Based Fake News Detection Using Ensemble Learning

“…The highest accuracy was obtained when using unigrams features and linear SVM, which gave an accuracy of 92%. This work is similar to that conducted in [ 22 ]; however, like previous authors, it ignores the strength of ensemble learning prediction in fake news detection.…”

“…Determines that Naïve Bayes is best for fake news detection Their work ignores the use of ensemble learning in fake news detection and only focuses on supervised learning approaches [ 12 ] Investigates ML methods for fake news detection. Determines the highest accuracy was obtained when using unigrams features and linear SVM This work is similar to that conducted in [ 22 ], however, like previous work, it ignores the strength of ensemble learning prediction in fake news detection [ 19 ] In their work, they propose a supervised learning-based technique for the detection of fake reviews from online textual content Their work only investigates the textual features of the news and utilises only supervised learning for prediction rather than ensemble learning [ 3 ] Their study explores different textual properties that can be used to distinguish fake content from real by training different models with the fake news features In their paper, they propose a solution to the fake news detection problem using the machine learning ensemble approach, however, they focus only on the textual analysis and ignore the importance of the publisher’s credibility [ 13 ] In their article, they proposed an ensemble classification model for the detection of fake news that has achieved a better accuracy compared to the state-of-the-art. The proposed model extracts key features from the fake news datasets, and the extracted features are then classified using the ensemble model consisting of three popular machine learning models namely, Decision Tree, Random Forest, and Extra Tree Classifier Their model does not use features that cover the entire spread of those related to fake news as well and the model ignores the combination of Boosted Decision Trees and Neural Networks [ 17 ] In their exploration, they found that among the different machine learning algorithms used, Gradient Boosting with optimized parameters performs the best for a multi-class fake news dataset Their research is an attempt to improve the existing fake news classification using a multi-class dataset with the motivation that it can be helpful for future researchers working in this area.…”

“…The ensemble model consists of the following classification models as described in [ 16 ] and they are chosen based on the experiments conducted by [ 22 ].…”

“…Two-Class Boosted Decision Tree (BDT) The Two-Class Boosted Decision Tree model has been proposed by Microsoft Azure Machine Learning Studio (classic). This paper utilises the Two-Class Boosted Decision Tree based on the results stated in [ 22 ] and further experimentation. According to [ 22 ], from the experiments performed, and the results obtained, the Two-Class Boosted Decision Tree performed the best.…”

“…This paper utilises the Two-Class Boosted Decision Tree based on the results stated in [ 22 ] and further experimentation. According to [ 22 ], from the experiments performed, and the results obtained, the Two-Class Boosted Decision Tree performed the best. Hence, it can be concluded that based on our selected dataset the Two-Class Boosted Decision Tree is the best method suited for detecting and predicting Credibility-Based Fake News.…”