trust-sense 0.1.0

Advanced Multimodal AI Library for Fake News & Psyops Detection

Trust Sense — Full AI Intelligence Platform

Production-ready, scalable AI platform for trust and credibility analysis of text, audio, video, and images.
Backend: Python 3.11 + FastAPI + SQLAlchemy + Claude/HuggingFace + WebSockets
Frontend: React 18 + Vite + TailwindCSS + Recharts
Security: JWT, RBAC (Admin/Analyst/Viewer), CORS, Rate Limiting, Network Policies
Deployment: Kubernetes with PostgreSQL, Nginx Ingress, SSL/TLS

📚 Documentation

• DEPLOYMENT.md — Production deployment guide (Kubernetes, security, monitoring)
• API Docs — Available at http://localhost:8000/docs when running locally

---

⚡ Quick Start (5 minutes)

1. Backend Setup

cd backend

# Create virtual environment
python -m venv venv

# Activate (Mac/Linux)
source venv/bin/activate

# Activate (Windows)
venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Configure .env with API keys (see below)
# nano .env  or  code .env

# Start the API server
python -m uvicorn app.main:app --reload --port 8000

Backend: http://localhost:8000
API Docs: http://localhost:8000/docs

2. Frontend Setup

cd frontend
npm install
npm run dev

Frontend: http://localhost:5173

3. Configure API Keys

Edit backend/.env:

# NLP Engines
ANTHROPIC_API_KEY=sk-ant-...     # https://console.anthropic.com
HUGGINGFACE_API_KEY=hf_...       # https://huggingface.co/settings/tokens

# Power BI Integration (Optional)
POWERBI_DATASET_ID=your-dataset-id
POWERBI_TOKEN=your-powerbi-token

---

🎯 Modules & Features

Text Analysis ✅

• Trust score (0-100)
• Sentiment analysis (-1 to +1)
• Fake news probability
• Manipulation detection
• Emotional signals
• Language analysis

curl -X POST http://localhost:8000/api/analyze-text \
  -H "Content-Type: application/json" \
  -d '{"text":"Some text to analyze"}'

Audio Analysis ✅

• Speech-to-text transcription
• Voice emotion detection (joy, sadness, anger, fear, neutral)
• Pitch & energy metrics
• Transcript credibility analysis

curl -X POST http://localhost:8000/api/analyze-audio \
  -F "file=@speech.mp3"

Video Analysis ✅

• Deepfake detection (FaceNet512 embeddings)
• Facial emotion analysis (7 emotions)
• Face consistency scoring
• Frame-by-frame emotion timeline

curl -X POST http://localhost:8000/api/analyze-video \
  -F "file=@video.mp4"

Multimodal Analysis ✅

• Combined audio + video analysis
• Cross-modal consistency checking
• Integrated risk assessment

curl -X POST http://localhost:8000/api/analyze-multimodal \
  -F "audio_file=@audio.mp3" \
  -F "video_file=@video.mp4"

Image Analysis ✅

• OCR (optical character recognition)
• Image description & understanding
• Visual credibility assessment

Power BI Integration ✅

• Real-time data sync
• Custom dashboards
• Visualization & reporting
• REST API integration

---

📁 Project Structure

trust-sense/
├── backend/
│   ├── app/
│   │   ├── main.py                      # FastAPI app entry
│   │   ├── core/
│   │   │   ├── config.py                # Settings (UPDATED)
│   │   │   ├── database.py              # SQLAlchemy
│   │   │   └── security.py              # JWT & auth
│   │   ├── models/
│   │   │   └── models.py                # DB models (UPDATED)
│   │   ├── routers/
│   │   │   ├── auth.py                  # Authentication
│   │   │   ├── analysis.py              # Text analysis
│   │   │   └── audio_video.py           # AV analysis (NEW)
│   │   └── services/
│   │       ├── nlp_service.py           # Claude+HuggingFace (ENHANCED)
│   │       ├── av_service.py            # Audio/Video (NEW)
│   │       └── powerbi_service.py       # Power BI (NEW)
│   ├── requirements.txt                 # ALL deps (UPDATED)
│   └── .env                             # Config (UPDATED)
└── frontend/
    └── src/
        ├── pages/
        ├── components/
        └── services/api.js

---

🔌 API Endpoints

Method	Endpoint	Description
POST	/auth/register	Register user
POST	/auth/login	Login, get JWT
POST	/api/analyze-text	Text analysis
POST	/api/analyze-audio	Audio analysis
POST	/api/analyze-video	Video analysis
POST	/api/analyze-multimodal	Combined A/V
GET	/api/powerbi-status	Power BI status
POST	/api/powerbi-sync	Manual sync
GET	/api/analysis-history	History (auth)
GET	/api/stats	Statistics (admin only)
GET	/api/stats/summary	Public summary stats
GET	/health	Health check
GET	/docs	Swagger UI
WebSocket	/ws/live	Live alerts

---

🧠 NLP Engine Features

Primary: Anthropic Claude

• Advanced reasoning
• Image analysis (with vision)
• Structured JSON output
• Multilingual support

Fallback: HuggingFace

• Distilbert sentiment analysis
• Local model caching
• No API key required
• Fast inference

Graceful Degradation

1. Try Claude API
2. If unavailable → Fall back to HuggingFace
3. If HuggingFace fails → Use rule-based analysis

---

📊 Power BI Integration

Real-time data sync from Trust Sense to Power BI dashboards.

Setup: Follow POWERBI_SETUP.md

Synced Data:

• Analysis ID, User ID, Input Type
• Trust Score, Sentiment, Credibility
• Risk Level, Emotions, Timestamps
• Deepfake Probability, Voice Emotion

Triggers:

• Automatic on analysis completion
• Manual via /api/powerbi-sync endpoint
• Batch sync via background jobs

---

🎓 Usage Examples

Python

import requests

# Text analysis
response = requests.post(
    "http://localhost:8000/api/analyze-text",
    json={"text": "Breaking news about AI safety"}
)
print(response.json()["trust_score"])

# Audio analysis
with open("speech.mp3", "rb") as f:
    response = requests.post(
        "http://localhost:8000/api/analyze-audio",
        files={"file": f}
    )
print(response.json()["voice_emotion"])

JavaScript/React

// Text analysis
const response = await fetch("http://localhost:8000/api/analyze-text", {
  method: "POST",
  headers: { "Content-Type": "application/json" },
  body: JSON.stringify({ text: "Some text..." })
});
const result = await response.json();
console.log(result.trust_score);

See QUICK_REFERENCE.md for more examples.

---

📚 Documentation

File	Purpose
PROJECT_DESCRIPTION.md	Full project description — vision, features, stack, API, security, deployment
PROJECT_ROADMAP.md	Tech roadmap, phases, free hosting link
HOSTING.md	Step-by-step free hosting (Vercel, Render, Upstash)
INTEGRATION_GUIDE.md	Setup & architecture (if present)
POWERBI_SETUP.md	Power BI configuration (if present)
QUICK_REFERENCE.md	Code examples (if present)

---

🔐 Security Features

• JWT authentication
• Password hashing (bcrypt)
• RBAC: Admin, Analyst, Viewer roles
• CORS (configurable via CORS_ORIGINS)
• Rate limiting by subscription tier
• Environment variable secrets

---

📈 Performance

• Text: < 500ms per analysis
• Audio: 2-5s (depends on file size)
• Video: 5-15s (depends on length)
• Batch sync: Bulk Power BI uploads

---

🚀 Deployment

Docker

docker compose up -d --build

Frontend: http://localhost:5173 — Backend: http://localhost:8000 — Redis: 6379

Free hosting (Vercel + Render)

See HOSTING.md for step-by-step free deployment.

Other

• Railway / Heroku: Connect repo; set root dir and env (see HOSTING.md).
• render.yaml included for Render Web Service (root directory: backend).

---

🔧 System Requirements

• Python 3.11+
• Node.js 16+
• 4GB RAM minimum
• Optional: FFmpeg, Tesseract OCR

---

⚙️ Environment Variables

# Database
DATABASE_URL=sqlite:///./trust_sense.db

# Security
SECRET_KEY=your-secret-key
ALGORITHM=HS256
ACCESS_TOKEN_EXPIRE_MINUTES=1440

# NLP
ANTHROPIC_API_KEY=sk-ant-...
HUGGINGFACE_API_KEY=hf_...

# Power BI
POWERBI_DATASET_ID=...
POWERBI_TOKEN=...
POWERBI_API_URL=https://api.powerbi.com/v1.0/myorg
POWERBI_TABLE_NAME=AnalysisResults

---

🐛 Troubleshooting

Issue	Solution
CORS errors	Verify backend running on :8000
"Analysis failed"	Check API key, restart backend
Video processing slow	Reduce video size, use GPU
Power BI sync fails	Verify token/dataset ID

See INTEGRATION_GUIDE.md for detailed fixes.

---

📋 Feature Checklist

• ✅ Text analysis with Claude + HuggingFace + psychological analysis
• ✅ Audio transcription & emotion detection
• ✅ Video deepfake detection & facial emotions
• ✅ Image OCR & analysis
• ✅ Social media import (multiple platforms)
• ✅ Power BI real-time sync (optional)
• ✅ JWT auth + RBAC (Admin / Analyst / Viewer)
• ✅ WebSockets (/ws/live) for live alerts
• ✅ SQLite / PostgreSQL + Redis (Docker)
• ✅ REST API with Swagger docs
• ✅ React + Vite + TailwindCSS frontend
• ✅ Docker + CI/CD (GitHub Actions)
• ✅ Free hosting guide (Vercel + Render)
• ✅ Production-ready code

---

📞 Support

• API Docs: http://localhost:8000/docs (Swagger)
• Claude: https://docs.anthropic.com
• HuggingFace: https://huggingface.co/docs
• Power BI: https://learn.microsoft.com/power-bi
• FastAPI: https://fastapi.tiangolo.com

---

Version: 2.0.0
Status: Production Ready ✅
Last Updated: March 2025