Author : Jorge Sanchez
Publisher :
ISBN 13 :
Total Pages : pages
Book Rating : 4.:/5 (125 download)
Book Synopsis Advance-Fee Scam Email Classification Using Machine Learning by : Jorge Sanchez
Download or read book Advance-Fee Scam Email Classification Using Machine Learning written by Jorge Sanchez and published by . This book was released on 2021 with total page pages. Available in PDF, EPUB and Kindle. Book excerpt: This study investigates the effectiveness of supervised machine learning algorithms at classifying Advance-Fee fraud email messages, also known as 419 scams. Advance-Fee scams occur when a victim pays money to someone in anticipation of receiving something of greater value in return and then receives little or nothing in return. These scams are commonly perpetrated over email and, depending on the skills of the scammer and the susceptibly of the victim, can be financially and emotionally devastating to victims. For this reason, it is important to develop systems that catch these malicious emails before they reach potential victims. In the past supervised machine learning models have been successful at classifying general spam, with simple text-based models, that only analyze email body text, showing up to 95% accuracy. In this study, five text-based models were developed using five supervised machine learning algorithms that have previously been effective in the field of spam classification: Naïve Bayes, Support Vector Machine, Multilayer Perceptron, Logistic Regression, and Random Forest. The models developed in this study were compared to models that target general spam through text-based analysis. Results showed improvements in classification accuracy of Advance-Fee scams over general spam for all the models tested. In the case of Logistic Regression, targeting Advance-Fee scam messages showed an accuracy score of 99.1 %, a more than 4% improvement over models targeting general spam. These findings show that Advance-Fee scam emails should be targeted using models that were specifically trained for such messages. These finding also imply that targeting specific types of spam may be more effective than targeting many types of spam when using text-based models.