Yandex Zen

Zen is an example of the implementation of a specialized artificial intelligence technology.[12][13]

To analyze the interests and preferences of users, Yandex uses information about sites that have been visited, as well as user-specified interests.[14]

The system analyzes the user's favorite sites and other behaviors with the aim of creating a unique model of the user's preferences.[15] With an increasing amount of data about the user, the system can offer the user more relevant and topical content,[15] including content from sources unfamiliar to the user.[16] Zen adapts to the changing interests of the user.[17] For example, if a user begins to read about architecture, content on this subject will appear in their content feed more often.[18][19]

Prediction of user interest by using collaborative filtering[10]

The service is available as part of the desktop and mobile versions of Yandex Browser[11] and in Yandex Launcher.[20][21]

Zen, Launcher and Browser belong to the “Discovery” technology category (services and apps that use artificial intelligence to adapt to a user).[22][23][24]

The technology that underlies Zen was adapted by Yandex and CERN for use in the Large Hadron Collider. It is used to provide in-depth analysis of the results of physics experiments taking place at the LHС.[25][26]

In 2017, Yandex announced the launch of a platform that allows companies and independent authors to publish media content (articles, photos, videos) directly to Zen.[27] The platform also allows popular authors to earn money by using micropayment channels and ads.[27][28][29] According to Zen Media Director Daniel Trabun, the company plans to pay up to $1 million to the best authors of publications in 2017.[27][30]

Prior to the launch of the platform, Zen feeds consisted only of publications selected from publicly available sources.[31]

In 1997, Yandex began research into natural language processing, machine learning and recommendation systems.[20] In 2009, the proprietary machine learning algorithm MatrixNet was developed by Yandex, becoming one of the key components that Zen functions on.[10] The first Yandex service to introduce the use of recommendation technology was Yandex.Music, which was launched in September 2014.[1] This technology was then implemented in Yandex.Market and Yandex.Radio.[1]

In June 2015, a beta version of Zen became available.[1][2][32] At first, the Zen content feed showed only content from the media, and the service was only available to the 5% of users of Yandex Browser on Android that had registered a Yandex account.[1] Prior to this, Zen was available in an experimental form on the webpage zen.yandex.ru.[10][33]

In the following months, other types of content were added to Zen, such as image galleries, articles, blogs, forums, videos from YouTube, etc.[11][10][34]

According to data from April 2017, Zen is available in more than 50 languages in more than 100 countries, including the United States, India and Brazil.[3][11][35]

The head of the Zen service is Dmitry Ivanov.[36]

Zen is one of Yandex's experimental business activities.[22][37] According to data from 2015, this segment earned the company almost half a billion rubles in revenue.[38] In Q2 2016, Zen and other experimental services brought in 153 million rubles in revenue to Yandex.[37][39] Monetization of the service is due to ad blocks built into the news feeds (in Russia, ads are served by Yandex.Direct).[11]

After the release of Zen, several major companies also announced personal recommendation services. In May 2016, Mail.Ru Group presented a similar project called Likemore, which offers users content from VK (VKontakte) social media groups.[40][41] In August 2016, Google launched a test version of a similar service that recommends news articles.[40][42][43] Apple and Facebook have also launched news services with similar functionality.[44]