CarderPlanet
Professional
The law opened the curtain: what is behind the recommendations of our favorite platforms?
Leading companies such as VK, Yandex, Ozon, Wildberries, Sbermarket, M. video-Eldorado, ivi, Start have published information on their websites about how they use recommendation algorithms and "sorcerers". This is a requirement of the new law on recommendation algorithms, which came into force on October 1, 2023. The law applies to social networks, online stores, video services, and other platforms.
Recommendation algorithms (including search engine wizards) help online services select and sort data that is shown to the user in response to their query. By law, services must publish rules for using algorithmic data. These rules should describe what data is collected, how it is organized and analyzed, and how it is used to target user requests. You must also specify the types and sources of data to target.
Owners of websites and services that use recommendation algorithms have started implementing the law, Roskomnadzor confirmed to Vedomosti.
Most major services, such as Yandex, VK, Wildberries, Ozon, M. video-Eldorado, ivi, Start, also told the publication that they comply with the requirements of the law and posted information on the use of recommendation technologies on their resources.
Roskomnadzor can check compliance with the law and request information about recommendation technologies from the service owner, as well as access to software and hardware tools for evaluating them, lawyers said. If the ILC finds violations, it will require them to be corrected within 10 days, and if this is not done, it will send another notification with a request to stop using recommendation technologies. If the service owner does not comply with the ILC requirement within a day after receiving the notification, the ILC will block the service via mobile operators.
Monitoring user activity
On the Yandex home page, in the drop-down menu below the user agreement, the "Recommendation rules" item has appeared, which explains how all ecosystem services work: search, Yandex.Music, Kinopoisk, Bookmate, Yandex.Taxi, Yandex.Shops, Yandex.Market and others. For example, Yandex analyzes the user's search query history, clicks on links in the search results and the time of return to the search page, information about the user's device, geolocation, cookies, the language of queries and pages in the search results.
Yandex. Music also takes into account likes and preferences settings – the mood of tracks, the degree of familiarity, the similarity to other content, and in Kinopoisk – the history of searching for titles, evaluating them, and adding them to the collection.
VK, in its September 30 video on the social network's home page, also talked about how the company uses recommendation technologies. Then the link leads to a page with more detailed information about how the resource uses recommendation technologies. The information is duplicated on the main page of the social network.
"Users can influence the fact and order of content display by raising or lowering their rating," the VK rules explain. The social network collects data about user actions on the site, such as likes, reposts, subscriptions, complaints, and hides from the feed.
Ozon has a section called "Working with Ozon algorithms" in the menu, and the Wildberries website has a notification about the use of recommendation technologies at the bottom of the page. Sbermarket has a "Recommendation Policy" tab. Ozon uses about 300 product parameters – characteristics, rating, and customer behavior (order, view, or add to cart), according to the rules. Wildberries collects data about the user's actions on the site, such as the user's provider, browser, and operating system, as well as the resolution and color of the screen on the device specified on the site.
At Sbermarket, creating recommendations involves five stages: collecting, organizing, applying machine learning methods, checking and displaying recommendations. The rules for using recommendation technologies in the Start service can be found in the legal documents section. It follows that an online movie theater takes into account the preferred genres (based on the content viewed and user actions), the frequency of viewing, and whether the user sees the content to the end. The ivi video service examines information about views, purchases, favorites, ratings, and feedback.
The new version of the law does not require services to disclose commercial details of recommendation algorithms, but obliges them to make them transparent to users, experts say.
Deputy Chairman of the State Duma Committee on Information Policy, Information Technologies and Communications Anton Gorelkin reminded that foreign services present on the Russian market are also fully required to comply with the new requirements of Russian legislation. However, YouTube and Google have not yet fully disclosed how the algorithms work.
Leading companies such as VK, Yandex, Ozon, Wildberries, Sbermarket, M. video-Eldorado, ivi, Start have published information on their websites about how they use recommendation algorithms and "sorcerers". This is a requirement of the new law on recommendation algorithms, which came into force on October 1, 2023. The law applies to social networks, online stores, video services, and other platforms.
Recommendation algorithms (including search engine wizards) help online services select and sort data that is shown to the user in response to their query. By law, services must publish rules for using algorithmic data. These rules should describe what data is collected, how it is organized and analyzed, and how it is used to target user requests. You must also specify the types and sources of data to target.
Owners of websites and services that use recommendation algorithms have started implementing the law, Roskomnadzor confirmed to Vedomosti.
Most major services, such as Yandex, VK, Wildberries, Ozon, M. video-Eldorado, ivi, Start, also told the publication that they comply with the requirements of the law and posted information on the use of recommendation technologies on their resources.
Roskomnadzor can check compliance with the law and request information about recommendation technologies from the service owner, as well as access to software and hardware tools for evaluating them, lawyers said. If the ILC finds violations, it will require them to be corrected within 10 days, and if this is not done, it will send another notification with a request to stop using recommendation technologies. If the service owner does not comply with the ILC requirement within a day after receiving the notification, the ILC will block the service via mobile operators.
Monitoring user activity
On the Yandex home page, in the drop-down menu below the user agreement, the "Recommendation rules" item has appeared, which explains how all ecosystem services work: search, Yandex.Music, Kinopoisk, Bookmate, Yandex.Taxi, Yandex.Shops, Yandex.Market and others. For example, Yandex analyzes the user's search query history, clicks on links in the search results and the time of return to the search page, information about the user's device, geolocation, cookies, the language of queries and pages in the search results.
Yandex. Music also takes into account likes and preferences settings – the mood of tracks, the degree of familiarity, the similarity to other content, and in Kinopoisk – the history of searching for titles, evaluating them, and adding them to the collection.
VK, in its September 30 video on the social network's home page, also talked about how the company uses recommendation technologies. Then the link leads to a page with more detailed information about how the resource uses recommendation technologies. The information is duplicated on the main page of the social network.
"Users can influence the fact and order of content display by raising or lowering their rating," the VK rules explain. The social network collects data about user actions on the site, such as likes, reposts, subscriptions, complaints, and hides from the feed.
Ozon has a section called "Working with Ozon algorithms" in the menu, and the Wildberries website has a notification about the use of recommendation technologies at the bottom of the page. Sbermarket has a "Recommendation Policy" tab. Ozon uses about 300 product parameters – characteristics, rating, and customer behavior (order, view, or add to cart), according to the rules. Wildberries collects data about the user's actions on the site, such as the user's provider, browser, and operating system, as well as the resolution and color of the screen on the device specified on the site.
At Sbermarket, creating recommendations involves five stages: collecting, organizing, applying machine learning methods, checking and displaying recommendations. The rules for using recommendation technologies in the Start service can be found in the legal documents section. It follows that an online movie theater takes into account the preferred genres (based on the content viewed and user actions), the frequency of viewing, and whether the user sees the content to the end. The ivi video service examines information about views, purchases, favorites, ratings, and feedback.
The new version of the law does not require services to disclose commercial details of recommendation algorithms, but obliges them to make them transparent to users, experts say.
Deputy Chairman of the State Duma Committee on Information Policy, Information Technologies and Communications Anton Gorelkin reminded that foreign services present on the Russian market are also fully required to comply with the new requirements of Russian legislation. However, YouTube and Google have not yet fully disclosed how the algorithms work.
