Title: «Between the Lines»

Format: A cultural and entertainment digital book media.

«Between the Lines» is a medium that helps people choose books based on their mood, interests, and life circumstances.

Unlike traditional book projects focused on rankings, literary premieres, and genres, «Between the Lines» invites readers to discover literature through emotion and personal needs.

For example, a user might search not just for a science fiction novel, but for a book that has the atmosphere of the movie Interstellar, a story for a quiet evening, or a work that offers a new perspective on growing up and relationships.

Target Audience: Young adults aged 18–30 who are interested in culture, film, music, visual aesthetics, and self-development. This group includes both active readers and individuals looking to return to reading.

Main Platforms:

Telegram — the primary platform for book selections, reviews, polls, and personalized recommendations.

VK — a secondary platform for short videos, visual cards, and audience acquisition.

Main Sections:

«Book for Your Mood» — selections based on emotional state.

«If You Liked This…» — recommendations linked to films, series, music, and games.

«One Book for the Evening» — short and accessible works.

«Book Diagnosis» — recommendations tailored to subscriber requests.

«Is It Worth Reading?» — brief reviews.

«Spoiler-Free Book» — descriptions of the work’s atmosphere, complexity, and emotional content.

The medium’s main goal is to make finding the right book simple and personalized, while maintaining a friendly and unobtrusive communication with the audience.

For the «Between the Lines» project, both external and internal data sources are utilized. These sources help in discovering books, generating recommendations, tracking reader interests, and assessing the effectiveness of publications.

1. Google Books APISource for bibliographic information about books.Data utilized:work title;author;publication year;genre and categories;description;page count;ISBN;cover image.This data is used to create book cards and form a unified book catalog.2. Open LibraryOpen bibliographic data database.Data utilized:author and work information;various editions of books;publication languages;work identifiers;covers and descriptions, if available.This source allows for supplementing Google Books information and verifying bibliographic details.3. Book Websites and Reading CommunitiesFor example, LiveLib and Goodreads.Data utilized:reader ratings;popular books;public reviews;thematic lists;discussed works.These sources help the editorial team understand reader interests and find topics for publications. Information is collected manually or through authorized technical means while adhering to platform rules.4. Cultural and Informational SourcesMovie posters, music platforms, cultural publications, and themed communities.Data utilized:popular films and series;cultural trends;aesthetic directions;topics discussed on social media.This information is primarily needed for the «If you liked…» section, which connects literature with other cultural phenomena.

1. Telegram Channelnumber of post views;reactions;comments;poll results;subscriber questions and requests;subscriber growth dynamics.2. VK Communitypost reach;clip views;reactions and comments;shares;audience engagement.3. Editorial DatabaseAn internal table containing:books previously mentioned in media;thematic categories;emotional characteristics of works;published materials;scheduled publications;editorial review results.4. Audience FeedbackComments, survey responses, and voluntarily submitted user messages.These help determine which topics genuinely interest readers and shape personalized recommendations.

Working with information is built on a hybrid model: automated services handle data searching, systematization, and preliminary analysis, while the editor is responsible for interpretation, the quality of recommendations, and final publication.

Machine Processing: Receiving bibliographic data via API; Consolidating information from multiple sources; Duplicate removal; Gathering accessible statistics from Telegram and VK; Structuring survey results.

Human Processing: Identifying compelling literary themes; Analyzing reader discussions; Verifying the accuracy of collected information; Selecting audience queries for future publications.

Result: A unified database of books, user requests, and statistics.

Each book receives a set of characteristics necessary for recommendations.

The connotations of mood, pacing, complexity, and cultural associations are editorial interpretations, not objective characteristics of the book.

Machine Processing:

• Book distribution by genre and theme;
• Preliminary determination of emotional characteristics based on descriptions;
• Similarity search between the user query and book records;
• Generation of a list of potential recommendations;
• Identification of recurring themes.

Human Processing:

• Verification of suggested tags;
• Refinement of the atmosphere and emotional complexity of the works;
• Assessment of how well the book matches a specific query;
• Exclusion of unsuccessful or unsuitable recommendations.

Result: A structured book catalog with verified characteristics.

Machine Processing:Searching for books based on set criteria;Preparing drafts for selections;Forming card templates;Adapting long text for short VK format;Checking for repetitions and technical errors.Human Processing:Selecting the main idea of the publication;Reviewing the content of recommendations;Writing and editing the final text;Adding authorial commentary;Checking for the absence of unwanted spoilers;Approving the visual design.Result: Finished editorial material.

Machine Processing:

Delayed publication of approved materials; Retrieving available statistics; Engagement counting; Comparing section performance; Preparing the weekly analytical report.

Human Processing:

Analyzing the reasons for publication success or failure; Evaluating comment quality; Adjusting the thematic plan; Making decisions about new formats.

Outcome: Updating the content strategy based on audience reaction.

Automated tools are used to accelerate editorial work, but they do not replace editorial responsibility.

The automated service must use verifiable sources and maintain records of data provenance.

Rules:

Do not create non-existent books, author names, or bibliographic information.

If data conflicts arise from different sources, submit the record for manual review.

Do not present automatically assigned emotional characteristics as objective facts.

If information is missing, designate it as unknown rather than fabricating it.

Do not include links to sources that were not actually used.

The primary objective of the automation is to find books that align with the user’s interests and emotional needs.

Rules:

Consider the mood, theme, pacing, and complexity of the work.

Do not rely solely on popularity or rating.

Do not recommend a book based only on genre matching.

Account for any topics the user has specified as undesirable.

Propose multiple options if the request allows for varied interpretations.

Explain why a specific book suits the request.

For example, if the request is «I want something similar to Interstellar, ” the system should search not just for books about space, but also for works dealing with time, scientific discovery, human relationships, and philosophical questions.

AI can be used to draft outlines, headlines, short descriptions, and adaptations of publications.

Style Guidelines:

Friendly, accessible language; No elitism; Moderate emotionality; Minimum reliance on clichés; Avoid definitive judgments.

Forbidden Phrasing:

«Everyone must read this book.» «The greatest work of all time.» «This book will guaranteed change your life.»

Preferred Phrasing:

«Try this book if you enjoy slow-paced stories.» «May interest those drawn to philosophical science fiction.» «It should be noted that the book is emotionally complex.»

All generated texts undergo editorial review before publication.