AI to Help Russian Publishers Predict Bestsellers

Can You Predict if a Book Will Sell?

Publishing continues to evolve with digital tools, and AI is now stepping into territory once considered untouchable: choosing which books are likely to succeed. Existing automated solutions already speed up editing and production. Now, artificial intelligence is being used not just to create and distribute books but to forecast potential bestsellers.

Maria Kuteynikova, a graduate of a joint master’s program between Tomsk State University and Skillfactory, under the guidance of Professor Zoya Rezanova, has developed a unique NLP-based model. It analyzes the “emotional curve” of literary works to predict their market success.

“The core idea is that a literary text can be seen not just as a sequence of events, but as a continuous flow of emotions guiding the reader from exposition to climax and resolution,” Kuteynikova explained.

Replacing the Reviewer

The Python-based model evaluates the emotional tone of each sentence using a language model tailored to Russian. Kuteynikova trained the system on a dataset of more than 4,000 books provided by publishers. Results revealed a clear link between a story’s emotional arc and its commercial outcome.

Artificial intelligence opens incredible possibilities for the publishing market, turning dreams into reality. Bringing AI into workflows—from page layouts to covers and text analysis—will transform the industry in the near future. Imagine a future where complex tasks are solved instantly, freeing time for creativity and innovation. AI will make books more engaging and accessible, improve quality, and simplify production. The publishing industry is on the brink of major change. I look forward to the day AI brings fresh ideas and solutions, opening new horizons for creative and functional publishing

Kir Vitkovsky

founder of the innovative layout service “Metranpazh” and design director

Unlike human reviewers, who require time to read and assess a manuscript, the AI works continuously and delivers insights far faster. Kuteynikova is now piloting the model at Eksmo, refining it based on editor feedback.

A Revolution in Publishing

Kuteynikova’s project marks the first time NLP analytics have been applied to Russia’s publishing market, underscoring a significant step in digitizing creative industries. The model is designed not just to streamline production, but to improve content selection itself—potentially making publishing more profitable.

Researchers worldwide have long studied emotional arcs. For example, a study of more than 1,700 works identified six archetypal story plots. But no one attempted to commercialize the insight. Kuteynikova’s system changes that by applying emotional analysis directly to the task of identifying books with guaranteed resonance.

Kuteynikova’s model shows how AI can transform the painstaking, subjective process of reviewing manuscripts into something data-driven, scalable, and potentially global. After trials at Eksmo, it could be adopted by other publishers and adapted to different languages for international markets.