TF-IDF classification based Multinomial NaïveBayes model for spam filtering
Alan Chávez · NORMA · 2020
Email spam, better known as unwanted email messages, is the practice of sending unsolicited electronic messages with different intentions, commonly commercial purposes, or trying to commit criminal actions. Despite the numerous anti-spam measures nowadays, spam still being a problem all over the internet due to the low-cost and high impact that represents elaborate a spam campaign. Many different solutions exist to categorize incoming messages such as white list, grey list, blacklist, Machine Learning, Rule-based filtering, etc. However, no one definitively. A possible reason is since spammers are high resilient, once a spam filtering method is compromised spammers adapt to it. The aim of the present work has the objective of detecting in a more effective way spam email with the Multinomial Naïve Bayes approach, in addition to text sanitation and TF-IDF. Results given by the proposed model gives an accuracy improve than Multinomial Naïve Bayes by its own.