In 2020, Yandex.Mail processed 78.3 billion emails, and a third of them were sent as spam.
Analysts of the service examined the filtered emails and found out which topics were actively used by spammers in the past year. Topped the mailing list about various kinds of payments – mainly related to the coronavirus.
Here’s what the top 5 topics used by spammers were :
- Compensation for coronavirus treatment costs
- Additional payments for those in need
- Lottery to support the population during self-isolation
- Compensation for the cost of imported goods under the import substitution program
- Demanding money for non-disclosure of personal information
At the same time, scammers often faked their letters to match messages from well-known sites such as Gosuslugi or mos.ru.
To protect users from spam, Yandex.Mail 360 uses Spam Defense technology, which uses complex algorithms and machine learning. First, it analyzes the basic factors: the sender’s address, text, links, attachments, etc., and then extracts more complex ones from them. The sum is about ten thousand factors. In particular, Yandex determines whether there have been complaints about letters from these addresses, whether an attachment has been sent to many people, and whether the text of the letter is similar to the text of a spam mailing.
A neural network model is used to detect the similarity of text to spam. Then “Spamoborona” processes the collected data using the CatBoost library and delivers its verdict. If the email is definitely spam, it will not be delivered. And if in doubt, Yandex will deliver it, but put it in the Spam folder so that the user can decide his fate himself.
Spam Défense algorithms are constantly being trained to recognize new types of spam. Users help a lot with this by marking unwanted emails as spam and reporting them to the Yandex.Mail 360 team.