Gravar e-mail: A Comparative Study of Unsupervised Machine Learning and Deep Learning Techniques for Anomaly Detection in Recommender Systems