1. Email Campaign Evaluation Based on User and Mail Server Response.
- Author
-
Szpyrka, Marcin, Suszalski, Piotr, Obara, Sebastian, and Nalepa, Grzegorz J.
- Subjects
EMAIL ,SPAM email ,PHISHING ,ARTIFICIAL neural networks - Abstract
The goal of an email service provider company is to send out a large number of emails to help its clients realise successful email marketing activities. Thousands of emails sent every minute need to be analysed in real time to reduce spam or phishing. The paper describes a method that uses real-time tracking of key campaign metrics such as the opens count, clicks count, hard bounces count, etc., to identify campaigns that should be stopped because they can be classified as spam or phishing. The key point of this solution is that we do not analyse email content. Nevertheless, the proposed neural networks are highly effective—the F1-score is above 0.95 for any used sample. Furthermore, the approach allows us to use the same model regardless of the language of an email. The method was developed and verified in collaboration with Freshmail, a leading provider of email marketing services in Poland. Validation of the method on real data provided by the company confirmed its high effectiveness. [ABSTRACT FROM AUTHOR]
- Published
- 2023
- Full Text
- View/download PDF