Skip to content

Latest commit

 

History

History
121 lines (92 loc) · 4.91 KB

README.md

File metadata and controls

121 lines (92 loc) · 4.91 KB

Guided Learning Module

Overview

Guided Learning is a personalized learning system based on notebook content. The system analyzes all records in the notebook, generates a progressive knowledge point learning plan, and helps users gradually master all content through interactive pages and intelligent Q&A.

Features

  1. Intelligent Knowledge Point Location (LocateAgent)

    • Analyzes all records in the notebook (solve, question, research, Co-Writer)
    • Identifies core knowledge points and organizes them in progressive relationships
    • Generates 3-5 structured knowledge point learning plans

  2. Learning Progress Management (GuideManager)

    • Tracks current learning state
    • Manages transitions between knowledge points (based on explicit user progression signals)
    • Provides learning progress feedback

  3. Interactive Page Generation (InteractiveAgent)

    • Converts knowledge points into visual, interactive HTML pages
    • Designs appropriate interactive elements based on knowledge characteristics
    • Supports HTML bug fixing functionality

  4. Intelligent Q&A Assistant (ChatAgent)

    • Answers user questions during learning
    • Provides contextually relevant answers based on current knowledge point and chat history
    • Provides additional explanations for potential user difficulties

  5. Learning Summary Generation (SummaryAgent)

    • Generates personalized learning summaries after completing all knowledge points
    • Analyzes learning process and mastery level
    • Provides follow-up learning suggestions

Directory Structure

guide/
├── __init__.py
├── guide_manager.py          # Session manager (includes learning progress management logic)
├── agents/
│   ├── __init__.py
│   ├── base_guide_agent.py   # Agent base class
│   ├── locate_agent.py        # Knowledge point location agent
│   ├── interactive_agent.py    # Interactive page generation agent
│   ├── chat_agent.py          # Q&A agent
│   └── summary_agent.py       # Summary generation agent
└── prompts/
    ├── zh/                    # Chinese prompts
    │   ├── locate_agent.yaml
    │   ├── interactive_agent.yaml
    │   ├── chat_agent.yaml
    │   └── summary_agent.yaml
    └── en/                    # English prompts (optional)

API Endpoints

REST API

  • POST /api/v1/guide/create_session - Create learning session
  • POST /api/v1/guide/start - Start learning
  • POST /api/v1/guide/next - Move to next knowledge point
  • POST /api/v1/guide/chat - Send chat message
  • POST /api/v1/guide/fix_html - Fix HTML page
  • GET /api/v1/guide/session/{session_id} - Get session information
  • GET /api/v1/guide/session/{session_id}/html - Get current HTML

WebSocket

  • WS /api/v1/guide/ws/{session_id} - Real-time interaction endpoint

Usage Flow

  1. Select Notebook

    • User selects a notebook containing records in the frontend
    • System calls create_session to create a learning session

  2. Generate Learning Plan

    • LocateAgent analyzes notebook content
    • Generates 3-5 progressive knowledge points
    • Displays learning plan to user

  3. Start Learning

    • User clicks "Start Learning"
    • System generates interactive page for first knowledge point
    • User can view interactive content on the right side

  4. Learning Interaction

    • User can ask questions in the left chat box
    • ChatAgent answers questions based on current knowledge point
    • User can click "Next" to move to next knowledge point

  5. Complete Learning

    • After completing all knowledge points, system generates learning summary
    • Summary includes learning review, mastery assessment, and improvement suggestions

Data Storage

All session data is stored in the user/guide/ directory, with each session saved as an independent JSON file:

  • File name format: session_{session_id}.json
  • Contains complete session state, knowledge points, chat history, etc.

Configuration Requirements

  • LLM environment variables must be configured (LLM_MODEL, LLM_BINDING_API_KEY, LLM_BINDING_HOST)
  • Ensure notebook has sufficient records (at least 1)
  • Recommend using LLM models that support JSON format output

Notes

  1. Knowledge Point Count: System automatically generates 3-5 knowledge points to ensure learning plan is neither too general nor too fragmented
  2. HTML Generation: If LLM-generated HTML has issues, can use Debug functionality to fix
  3. Chat History: Each knowledge point's chat history is independent for easier context management
  4. Session Persistence: Session data is automatically saved, can resume learning progress at any time

Extensibility

  • Can add more interaction types (e.g., animations, 3D visualizations)
  • Can integrate more learning tools (e.g., practice questions, quizzes)
  • Can add learning path recommendation functionality
  • Can support multi-user collaborative learning