E-Contenta algorithm can predict what you would love to read without even knowing you

ST. PETERSBURG, Russian Federation - March 22, 2015 - PRLog -- Personalized recommendations help even the visitor, who comes to the site for the first time, to see the information that corresponds to his (her) interests. During repeated visits the recommendations become even better, because the machine learning technology allows to remember the history of purchases or pages review and to adjust the content depending on the behavior at the site.

«Part of new users comes to the service without knowing, what to read. In this pilot project we decided to focus exactly upon them, — says Konstantin Savenkov, operation director of Bookmate company. — It was necessary to solve the problem of the cold start for personal recommendations, because at the moment of registration practically nothing is known about a new user. Guys succeeded, their algorithm works much better than non-personalized recommendations».

Observation of Bookmate service users showed: person that looks for the specific book, easily finds in Bookmate library and becomes a client. Some visitors begin to search for books in catalogue or address recommendations, generated by editors. However, writers' selections not always answer the expectations. Start-up successfully checked and confirmed the hypothesis: personal recommendations that consider the behavior of user in the net increases the number of clients that pay among the undecided readers.

At the present moment the E-Contenta algorithm builds the links between 500 thousand books and 1.5 mil of Bookmate users. Minimum time of service response does not exceed 30 milliseconds. Data about Vkontakte users subscriptions render special influence upon the algorithm quality. Today more than 280 thousand publics are analyzed. New users that did not make a decision about book selection open books, recommended by the algorithm, 2.17 times more frequently than the editors selections, conversion of such users into subscribers is 1.4 times higher.

Problem of user's monetization in the net has a key significance. By PWC evaluation media gains from paying on-line clients in comparison to off-line will reach only 17% in 2018 (10% in 2013).

«We plan to create maximally efficient recommendation product for media companies (press, audio, video and book services), — says Director General of E-Contenta Zoya Nikitina. — In spite of plenty of free pirate books, video and audio in Internet, E-Contenta will help the sellers of legal content to increase their profit».

About Bookmate

Bookmate service for searching for books and reading them is used monthly by 1.5 mil of readers, most of which are citizens of Russia, Ukraine, Kazakhstan and other Russian-speaking countries. In average a user spends 54 minutes a day, reading books in Bookmate. At the present moment the company's library contains more than 500 000 books from 300 publishers from all over the world. In the nearest plans of the company there is the expansion into the market of Turkey, Scandinavia and Latin America. Bookmate is a part of Dream Industries holding, which develops also such innovation products as Zvooq, Exchang.es, Theories and Practicies, 3^rd Place, Telegraph, Unisound.

About E-Contenta company
E-Contenta is a SaaS-service of personal recommendations for media-companies. The technology is capable to accumulate information about users behavior at on-line resource, as well as to process the external data from VKontakte and Facebook and other social networks, DMP (data management platform).