AI Learning Brain 🧠

A personalized tutoring system that transforms technical PDF textbooks into an adaptive, graph-based learning experience. Upload a textbook chapter, and the system builds a knowledge graph, plans a learning path tailored to what you already know, teaches each concept, and quizzes you before moving forward. Powered by LangGraph, Groq LLM, Neo4j, and sentence-transformers.

🚀 Overview

The AI Learning Brain is designed to solve the problem of information overload in technical subjects. It processes complex PDF documents, identifies core technical concepts, builds a hierarchical knowledge graph, and provides a multi-agent tutoring experience that adapts to the student's mastery level.

🏗 System Architecture

The project follows a modular, service-oriented architecture:

• Frontend: Streamlit provides a premium, responsive UI for both content processing and interactive learning.
• Orchestration: LangGraph manages the agentic workflow using a robust Command-based routing pattern.
• Brain (Knowledge Base): Neo4j stores the semantic relationships between concepts (Prerequisites, Subtopics, etc.).
• Session Management: TinyDB persists student progress, mastery scores, and historical learning paths.
• LLM: Groq (Llama-3.3-70B) powers the high-speed extraction and pedagogical reasoning.

🔄 End-to-End Workflow

1. Content Processing (The "Ingestion" Phase)

• PDF Extraction: Converts textbooks to clean Markdown using PyMuPDF4LLM.
• Intelligent Chunking: Splits text into technical segments while maintaining context.
• Structural Extraction: LLM analyzes each chunk to find technical concepts, difficulty scores, and LaTeX equations.
• Validation: Strict schema enforcement ensures zero data corruption before storage.

2. Brain Generation

• Knowledge Ingestion: The extracted JSON data is serialized and loaded into the Neo4j graph.
• Relationship Building: Automatically connects concepts based on prerequisites and shared units.

3. Learning Session (The "Tutor" Phase)

• Intent Parsing: Interprets student requests (e.g., "I want to learn about Op-Amps").
• Path Planning: Uses the MSMS (Multi-Source Multi-Sink) algorithm to find the mathematically optimal learning path based on the student's current mastery.
• Diagnostic Assessment: Generates targeted questions to evaluate the student's understanding.
• Remediation: If a student fails to answer correctly, the tutor provides a concise explanation addressing specific misconceptions and offers 3 attempts before adjusting the path.

📊 Visual Workflow

🛠 Setup & Installation

Prerequisites

• Python 3.10+
• Neo4j Database (Local or AuraDB)
• Groq API Key

Installation Steps

1. Clone the repository:

   git clone <repository-url>
   cd notebook-lm-mini

2. Setup Virtual Environment:

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate
   pip install -r requirements.txt

3. Configure Environment: Create a .env file in the root directory:

   GROQ_API_KEY=your_groq_key
   NEO4J_URI=bolt://localhost:7687
   NEO4J_USERNAME=neo4j
   NEO4J_PASSWORD=your_password

4. Run the Application:

   streamlit run app.py

📂 Project Structure

• app.py: Main Streamlit application entry point.
• src/services/: Core logic (PDF, LLM, Graph, Tutor, Planner).
• src/models/: Pydantic and TypedDict schemas for data consistency.
• src/database/: Database connection and persistence layers.
• data/: Temporary storage for extracted concept JSONs.
• sessions/: Persistent storage for student session data.

---

Built with ❤️ for Technical Education.