Booktube is a generative AI-powered web platform that transforms traditional textbooks and syllabi into interactive, bite-sized articles tailored to a learner’s style, preferences, and skill level. Built as part of a Human-Centered Design project, Booktube bridges the gap between bulky academic material and modern, personalized learning methods.
- 📄 PDF to Topic-wise Articles: Upload a syllabus and textbook PDF to get structured, easy-to-understand content.
- 🧠 Personalized Content Generation: Uses a RAG (Retrieval-Augmented Generation) pipeline powered by Gemini 2.0 Flash to generate articles tailored to user inputs like learning style, skill level, and prompts.
- 🎯 Recommendation Engine: Keeps learners engaged by suggesting relevant articles and topics based on learning patterns.
- 🗂️ Playlists & Read Later: Save articles to custom playlists, mark them to read later, and keep track of learning history.
- 🖼️ AI-Picked Thumbnails: Dynamically selected visuals from Pexels/Unsplash using Gemini for visually appealing article thumbnails.
- 📈 Progress Tracking: Monitor learning progress across chapters and subjects.
- Frontend: React.js, Tailwind CSS, Bootstrap
- Backend/AI: Gemini 2.0 Flash API, RAG Pipeline, Python (PDF Parsing)
- APIs & Libraries: Unstructured, Unsplash API, Pexels API, Google Cloud Platform
Booktube's architecture is designed to transform raw educational content into a personalized learning experience. The system follows a multi-stage pipeline, from content ingestion to personalized delivery.
-
Content Ingestion:
- The user uploads a textbook (PDF) and a syllabus via the React-based frontend.
- The backend, powered by Flask, receives these files.
-
Processing & Chunking:
- The uploaded PDF is parsed using the
Unstructuredlibrary to extract raw text and structure. - The content is then segmented into meaningful, topic-wise chunks based on the syllabus.
- The uploaded PDF is parsed using the
-
Personalized Content Generation (RAG Pipeline):
- When a user requests an article on a topic, the system uses a Retrieval-Augmented Generation (RAG) pipeline.
- Retrieval: Relevant chunks of the textbook are retrieved based on the topic and user's learning preferences.
- Augmentation: The retrieved context is combined with the user's prompt (e.g., "explain this like I'm a beginner").
- Generation: The augmented prompt is sent to the Gemini 2.0 Flash API, which generates a tailored, easy-to-digest article.
-
Engagement & Personalization:
- AI Thumbnails: Gemini is used to generate relevant search queries for Pexels/Unsplash APIs to fetch and assign visually appealing thumbnails to each article.
- Recommendation Engine: The system tracks user activity and suggests related articles to create a continuous learning path.
- User Features: Users can save articles to playlists, mark them for later, and track their progress, with all data managed by the backend.
Home Page
Generated Article View
