Clients of “Tinkoff” will be able to get feedback from the ecosystem about themselves, the press service of the bank said. Reviews will be based on their personal spending and the findings of the neural network.
Tinkoff will use a neural network to analyze customer spending and tell you what it thinks about themDenis Savosin
A neural network trained on customer reviews about Tinkoff products and services over 15 years of the company’s operation will generate similar reviews about them.
All testimonials will be styled after real testimonials from Tinkoff clients, which they have ever left on various online platforms and banking forums. Each review is formed individually, based on the user’s transactions.
The feedback generator was prepared in 3 stages:
- Data on spending was aggregated in categories by amount and number of purchases, by favorite stores and goods. As a result, more than 20 million records were collected, on the basis of which the first part of the review is collected;
- More than 20 thousand reviews were selected on specialized channels and websites. A neural network was trained on them so that the generated phrases looked like real customer reviews;
- Each review is structurally divided into two parts. The first is about customer spending, uses pre-recorded messages. The second – from the neural network – is always unique.
Tinkoff will tell you what it “thinks about customers” within the framework of a special project launched on the eve of its 15th anniversary. To participate, you need to leave feedback about Tinkoff’s work on one of five sites: Sravn.ru, Facebook, Instagram, VKontakte or Twitter, and also log in to project website until October 24, 2021.
Subscribe to our TG channelto keep abreast of all the news and events!
Cover photo: Nikita Burdenkov /
Source: RB.RU by rb.ru.
*The article has been translated based on the content of RB.RU by rb.ru. If there is any problem regarding the content, copyright, please leave a report below the article. We will try to process as quickly as possible to protect the rights of the author. Thank you very much!
*We just want readers to access information more quickly and easily with other multilingual content, instead of information only available in a certain language.
*We always respect the copyright of the content of the author and always include the original link of the source article.If the author disagrees, just leave the report below the article, the article will be edited or deleted at the request of the author. Thanks very much! Best regards!