Chubhi AI — Multi-Dimensional Semantic Analysis Ecosystem

Project Overview

Chubhi AI is a React + Vite Web Application that visualizes semantic relationships within the Guru Granth Sahib (via BaniDB API) using Hugging Face 3D vector embeddings and Plotly.js. This platform enables multi-dimensional exploration of sacred scripture through interactive visual analytics and semantic projections.

---

Core Application Features

Interactive 3D/2D Projections

Hardware-accelerated 3D vector space and three 2D projection planes (Meaning, Emotion, Topic) for comprehensive semantic visualization.

Dynamic Origin Translation

Clicking any Shabad dynamically translates coordinate origins $(0,0,0)$ with radial vector connection spokes, enabling perspective-based semantic analysis.

Bilingual Search & Gurmukhi Keyboard

Full-text search in English and Gurmukhi with an integrated on-screen Punjabi virtual keyboard for accessibility across language barriers.

Context Expansion ('Add Near')

Fetches surrounding sequential scriptural verses $(N-1, N, N+1)$ for localized comparative analysis and contextual understanding.

Random Discovery Engine

Randomly samples across 5,000+ Shabads in the corpus, facilitating serendipitous discovery and exploratory learning.

NLP Header Cleanup

Automatically strips metadata headers (Raag, Mehla, Slok) to score pure scripture content, ensuring accurate semantic embeddings.

Live Node Pruning

Interactive card filtering system allowing real-time manipulation of the visualization dataset and focus on specific semantic clusters.

Dark Mode

High-contrast dark UI optimized for extended reading sessions and reduced eye strain during deep study.

---

Environment Variables

The following environment variable is required at build time:

VITE_HF_TOKEN=<your-huggingface-api-token>

This token enables access to the Hugging Face Inference API for generating semantic embeddings.

Deployment Steps

1. Install Dependencies:

   npm install

2. Set Environment Variable: Configure VITE_HF_TOKEN in your Azure App Service configuration or build pipeline.

3. Build Application:

   npm run build

4. Deploy Static Assets: Upload the contents of the dist/ directory to your Azure Static Web App or App Service.

Azure Configuration

• Runtime: Node.js (for build process)
• Output Location: /dist
• API Routes: Not required (API calls are client-side to BaniDB and Hugging Face)
• CORS: Ensure BaniDB API and Hugging Face endpoints are accessible

---

Technical Stack

• Frontend: Vite
• Visualization: Plotly.js (WebGL-accelerated)
• NLP: Hugging Face Transformers API
• Data Source: BaniDB API (Guru Granth Sahib corpus)
• Styling: Modern CSS with dark mode support

---

License

This project is developed for educational and spiritual study purposes.