Moscow Polytech presented an efficient method of electronic mail protection from unsolicited messages

At the II International scientific and practical workshop “Society Digitalization: Current Conditions and Prospects” held at Moscow Polytech, Sherzod Khamidov a 3rd-year doctoral student at the “Cryptology” department of the Tashkent University of Information Technologies named after Muhammad Al-Khorazmiy, presented an innovative ensemble method for increasing the efficiency of protecting mail services from unsolicited messages.

Email is one of the most widely used means of communication in the modern world. In addition to exchange of messages, it is used for data storage, synchronizing devices and creating accounts on various online-services, including payment systems and online banking. However, this popularity also attracts attention of cybercriminals, who dispatch unsolicited and malicious emails in pursue of theft of confidential information and gaining access to users accounts.

The ensemble approach in machine learning is a combination of several models of classification for improving accuracy and reliability of results. In case of protection of emails from spam and malicious emails, the ensemble method allows using the advantages of various classification algorithms, compensating their drawbacks and improving the overall efficiency of the system.

Unlike traditional protection methods, such as keyword filtering or senders blacklists, the ensemble approach proposed by Sherzod Khamidov takes into account multiple characteristics of emails, including the structure and content of a text, metadata and behavioral patterns of senders. This allows for more accurate identification of unwanted messages and minimizes the number of false alarms.

Special attention is paid to preliminary processing of text data, which includes tokenization (splitting a text onto separate words and tokens), deleting stop-words (most common and meaningless words), and stemming (bringing words to their base) and vectorization (representing texts as numeric vectors for machine learning).

“Our ensemble method proved high efficiency for classifying electronic mails, surpassing existing solutions in accuracy and data processing velocity. It can be easily integrated into email services and anti-spam systems, providing reliable users protection from fishing attacks, theft of personal data and financial losses,” - noted Sherzod Khamidov

Reference: Moscow Polytech organized the workshop jointly with the Grozny State Petroleum Technical University named after Academician M. D. Millionshchikov, the Tashkent University of Information Technologies named after Muhammad Al-Khorazmiy, and “The Center for the Development of Digital Education" under the Ministry of Digital Technologies of the Republic of Uzbekistan. Head of the workshop Professor Yu.N. Fillipovich underlined the importance of such events for strengthening scientific cooperation between the countries.