cognitree

Description:

Cognitree is a web-based application that helps users explore complex topics visually. Each branch in the tree represents a unique piece of information retrieved using AI-powered search. Users actively grow, prune, and study branches to shape their learning path.

What it does:

• Generates structured knowledge areas from a topic
• Expands branches using real-time AI-powered search
• Prevents duplicate information using negative prompting
• Allows pruning and repositioning of branches
• Creates flashcards from retrieved content
• Creates quiz for self assessment
• Ask doubts using an AI chatbot with branch-level context
• Saves and restores complete game sessions (Coming Soon)

How it works (AI logic):

• Initial Knowledge Generation

POST /api/search
{
  "query": "machine learning"
}

• Gemini returns:

[
  "Supervised vs Unsupervised Learning",
  "Neural Networks",
  "Model Evaluation Metrics",
  "Overfitting and Regularization",
  "Real-world Applications"
]

These become the initial tree branches.

• Branch Growth with Non-Redundant Search

POST /api/web-search
{
  "query": "machine learning",
  "exclude": [
    "Supervised vs Unsupervised Learning",
    "Neural Networks"
  ]
}

The backend constructs:

"machine learning -supervised -neural networks"

Result:
New, unseen information
Attached to the selected branch
No repeated content across sessions

• Flashcard Generation

POST /api/create-flashcards
{
  "nodeId": "branch_12",
  "content": "Regularization helps prevent overfitting..."
}

Gemini returns structured flashcards linked to that node.

• Quiz Generation

POST /api/create-quiz
{
  "nodeId": "branch_12",
  "content": "Regularization helps prevent overfitting..."
  "numQuestions": 5
}

Gemini generates structured quiz questions (MCQ) linked to the selected node for active recall and self-assessment.

• AI Chatbot (Doubt Clarification)

POST /api/chat
{
  "nodeId": "branch_12",
  "question": "Why does regularization reduce overfitting?"
}

Gemini returns a focused, context-aware explanation based on the existing tree knowledge.

This allows users to:
Ask follow-up doubts while learning
Get explanations without leaving the tree
Keep answers aligned with their current learning path

Play Guide (Quick):

1. Enter a topic to generate the initial knowledge tree
2. Grow branches to explore new information
3. Prune branches that aren’t useful
4. Create flashcards from branches for revision
5. Add flowers/fruits to mark progress
6. Generate quiz for active recall and self-assessment
7. Ask doubts using the built-in AI chatbot
8. Save your tree and continue later (Coming Soon)

Why it’s useful:

• Reduces information overload
• Ensures every expansion is unique
• Encourages structured exploration
• Combines discovery, clarification, and revision in one interface
• Useful for research, self-study, and concept mapping

Tech Stack:

• Backend: FastAPI, SQLAlchemy, SQLite
• Frontend: Vanilla JavaScript, HTML5 Canvas
• AI: Gemini Chat + Search APIs
• Persistence: Public game state storage (Coming Soon)