Authorship attribution in twitter: a comparative study of machine learning and deep learning approaches

As social media platforms gain popularity and influence, content integrity and user accountability issues become more critical. Authorship attribution (AA) is a powerful tool for tackling such issues by accurately determining the real author of online posts. This study proposes an AA approach using machine and deep learning algorithms to accurately predict the author of unknown posts on social media platforms. It introduces Temporal Convolutional Networks (TCN) for short texts, investigates the effectiveness of combining Long Short-Term Memory (LSTM) and Convolutional Neural Networks (CNN), and explores the use of an Autoencoder combined with Adaboost classifier. This approach was tested on a Twitter dataset, achieving 52.77% accuracy in AA through multiple experiments across various scenarios.