A Deep Learning-Based Approach for Detecting Bangla Spam Emails
Riyadil Zannat, Afnan Alauddin Mumu, Abedur Rahman Khan, Tasneem Mubashshira, Sharifa Rania Mahmud · 2023
The field of wireless communications is expanding quickly in this 21st century. One of the most important communication tools available today is Email. Billions of emails are sent everyday throughout the world, that make people more vulnerable to dangers. Spam emails can be used for phishing, blackmail, and extortion from our digital equipment. Phishing emails are those that beg recipients for personal or financial information through different advertisements and offers from any e-commerce websites. Such spam emails are quickly identified by spam filters, which must work harder to block emails with no links or with few links. Despite multiple studies, some valid emails are still labeled as phishing and vice versa by spam filters. For recognizing spam emails in key languages including English, there are a variety of techniques. Nevertheless there aren’t many of these spam email detection tools accessible in Bangla. This study aims to categorize emails in Bangla language using deep learning approaches. In this research, a number of algorithms are investigated, additionally, we have developed a dataset and it is eventually concluded that the Bi-LSTM (Bidirectional Long Short-Term Memory) method has the greatest accuracy (97%) for the identification of Bangla phishing mail.