Using and Comparison of Artificial Intelligence Techniques to Detect Misinformation and Disinformation on Twitter

Abstract

This research investigates diverse artificial intelligence (AI) techniques for detecting misinformation on Twitter, addressing the pervasive concern of misinformation and fake news affecting public discourse. Employing models such as Long Short-Term Memory (LSTM), Support Vector Machine (SVM), Random Forest Classifier, Multinomial Naive Bayes and Gradient Boosting Classifier, we discern deceptive content from reliable information. Utilizing a dataset of 23,481 false tweets and approximately 21,417 real tweets, our analysis leverages Natural Language Processing (NLP), Deep Learning (DL) and Machine Learning (ML) techniques, showcasing the effectiveness of each model in identifying misinformation patterns. Our investigation rigorously assesses the strengths and limitations of AI techniques, focusing on accuracy, efficiency and scalability. Notably, the best results are achieved by models such as LSTM (98.84% accuracy, 98.79% F1 score), SVM (99.44% accuracy, 99.44% F1 score) and XGBoost Classifier (99.82% accuracy, 99.81% F1 score). The findings provide valuable insights into the performance of key models and serve as a resource for academics and researchers in the fields of artificial intelligence and social media analysis. Additionally, they provide practical guidance for supporting information integrity on Twitter, contributing to ongoing efforts to combat misinformation and enhance information credibility.