Résumé:
As social media platforms continue to grow in popularity and influence, the need to address
problems related to content integrity and user accountability becomes more critical. Authorship
attribution serves as a powerful tool in tackling such issues by accurately determining the
real author of online posts.
In this dissertation we propose an approach that utilizes deep learning models including
Temporal Convolutional Networks (TCN) and Long Short-term Memory Networks (LSTM)
combined with Convolutional neural networks (CNN) , along with machine learning models
such as Autoencoder and Adaboost to effectively predict the authors of unknown online posts.
To evaluate the effectiveness of our approach, we conducted experiments on various scenarios
using a Twitter dataset, where we achieved an accuracy rate of 52.77% in authorship attribution.
Keywords: NLP, Deep Learning, Authorship Attribution.
