VNSED: Vietnamese spam email detection using multi deep learning models

Email is one of the most popular communication methods today. However, a high percentage of spam emails are used for various purposes. Therefore, detecting spam emails and proposing solutions to limit spam are necessary. There are many current studies related to spam detection. Deep learning models have been utilized in numerous related studies to detect spam and achieve high accuracy. However, these deep learning models are mostly trained on English datasets. Adding test datasets for Vietnamese spam emails is essential to build spam detection models not only in English but also in Vietnamese. This study presents the construction of a Vietnamese spam email dataset and the proposed system named VNSED, which uses deep learning models including CNN (Convolutional Neural Network), BiLSTM (Bidirectional Long Short-Term Memory), and PhoBERT to detect Vietnamese spam email. Experimental results show that these deep learning models all achieve high accuracy in detecting Vietnamese spam emails. Specifically, the accuracy of the models are CNN: 88.42%, BiLSTM: 83.03%, and PhoBERT: 86.47%, respectively.