⚡ Nexora Intelligence Engine — Advanced Real-Time Edition

Transform news feeds into actionable strategic intelligence with AI-powered reasoning, sentiment analysis, and automated alerts.

🚀 Quick Start

# Install dependencies
pip install -r requirements.txt

# Run the application
streamlit run app.py

Application opens at http://localhost:8501

No API keys required — Uses public Google News RSS feeds.

✨ What's New in This Version

🎨 Modern Dark UI

Gradient cards with animations
Real-time AI signal monitor
Urgency badges with blink effects
Professional data visualizations

🔴 Enhanced Flash News

Real-time RSS feeds from multiple sources
Sentiment timeline charts
Keyword highlighting in headlines
Top movers tracking (positive/negative)
Smart alerts with urgency levels

🧠 Advanced Intelligence Brief

Strategic reasoning with Flan-T5
30-day sentiment forecasts
Scenario tree with probabilities (Bull/Base/Bear)
Action recommendations with impact scores
Confidence assessment with visual bars
Multi-format exports (MD/HTML/Playbook)

📈 Trends & Analytics

Dual-chart layouts (catalysts + momentum)
Volatility Index calculation
Sentiment evolution tracking
Entity momentum analysis
Burst activity detection

🔬 Research Lab

Topic clustering (K-Means)
Interactive network graphs (PyVis)
Source distribution heatmaps
Emerging themes discovery

🚨 Smart Alerts

Auto-pilot continuous monitoring
Urgency calculation (High/Medium/Low)
Correlation analysis (co-occurring catalysts)
Watchlist management
Alert history archiving

🧾 History & Exports

Session history with diff markers
ZIP export bundles (MD/HTML/JSON)
Timestamped filenames
Metadata tracking

📊 Key Features

Feature	Description
Real-Time Feeds	Google News RSS (Business, Technology, Global)
Sentiment Analysis	VADER-based scoring (-1 to +1)
Catalyst Detection	8 categories (M&A, Earnings, Regulatory, etc.)
AI Reasoning	Strategic briefs using Flan-T5 models
Network Graphs	Entity-Catalyst relationship mapping
Auto-Pilot Mode	Hands-free continuous monitoring
Caching	5-minute TTL for performance
No API Keys	Completely free to use

🎯 Use Cases

Market Intelligence — Track sector trends and sentiment shifts
Risk Monitoring — Automated alerts for negative catalysts
Strategic Planning — AI-generated scenario analysis
Competitive Analysis — Entity tracking and momentum detection
Research — Topic clustering and correlation discovery

📐 Architecture

Nexora Intelligence Engine
├── Real-Time RSS Feeds (Google News)
│   ├── Sentiment Analysis (VADER)
│   ├── Catalyst Classification (8 categories)
│   └── Credibility Scoring (domain-based)
│
├── AI Reasoning Layer
│   ├── Article Summarization (BART/Flan-T5)
│   ├── Strategic Analysis (Flan-T5)
│   └── Scenario Generation (Bull/Base/Bear)
│
├── Analytics Engine
│   ├── Trend Analysis (momentum, volatility)
│   ├── Topic Clustering (K-Means)
│   ├── Network Graphs (NetworkX + PyVis)
│   └── Correlation Detection
│
├── Alert System
│   ├── Keyword Monitoring
│   ├── Urgency Calculation
│   ├── Auto-Pilot Mode
│   └── History Archiving
│
└── Storage & Caching
    ├── FAISS (vector search)
    ├── SQLite (article storage)
    └── JSON (watchlists, alerts)

🎨 Screenshots

Flash News Dashboard

Live sentiment timeline
Top movers (positive/negative)
Smart alerts with urgency badges

Intelligence Brief

Strategic scenarios with probabilities
30-day sentiment forecast
Action cards with impact scores

Trends & Timeline

Dual-chart momentum analysis
Sentiment evolution over time
Volatility index tracking

Research Lab

Interactive network graphs
Topic clustering visualization
Source distribution heatmaps

📚 Documentation

Full Documentation — Complete feature guide
Test Suite — Core functionality tests

🔧 Configuration

Environment Variables (Optional)

# Use smaller models for faster loading
export NEXORA_STRATEGY_MODEL="google/flan-t5-small"
export NEXORA_SUMMARY_MODEL="sshleifer/distilbart-cnn-12-6"
export NEXORA_EMBEDDER="sentence-transformers/all-MiniLM-L6-v2"

# Adjust log level
export NEXORA_LOG_LEVEL="INFO"  # DEBUG, INFO, WARNING, ERROR

Sidebar Controls

Focus Topic — Research area (e.g., "AI chips market")
News to Fetch — 5-50 articles per refresh
Historical Lookback — 7-90 days for context
Auto-refresh — 5-60 seconds interval
Alert Keywords — Comma-separated (e.g., "NVIDIA, AMD")
Auto-Pilot Mode — Continuous monitoring toggle

🧪 Testing

Run the test suite to verify core functionality:

python3 test_core.py

Expected output:

✅ PASS — Imports
✅ PASS — Utilities
✅ PASS — Sentiment
✅ PASS — Catalysts
✅ PASS — Alerts
✅ PASS — Trends
✅ PASS — Exports

🎉 ALL TESTS PASSED — System ready for deployment

📦 Dependencies

Core

streamlit==1.39.0 — UI framework
aiohttp==3.10.5 — Async HTTP
feedparser==6.0.11 — RSS parsing

ML/AI

transformers==4.45.2 — LLM reasoning
sentence-transformers==3.0.1 — Embeddings
torch>=2.2.0 — PyTorch backend
vaderSentiment==3.3.2 — Sentiment analysis

Data/Viz

pandas==2.2.3 — Data manipulation
altair==5.5.0 — Charts
plotly==5.24.1 — Interactive plots
networkx==3.2.1 — Graph algorithms
pyvis==0.3.2 — Network visualization

Storage

faiss-cpu==1.12.0 — Vector search
scikit-learn==1.7.2 — Clustering

See requirements.txt for complete list.

🚀 Performance

Startup Time: ~10s (first run with model downloads)
Refresh Cycle: ~2-3s (25 articles)
Memory Usage: ~1.5GB (with models loaded)
Cache Hit Rate: ~80% (5min TTL)
Alert Latency: <1s (keyword matching)

🔐 Privacy & Security

✅ No API keys required
✅ Public RSS feeds only
✅ Local model execution
✅ Data stored locally
✅ No tracking or analytics
✅ Open-source and auditable

🛠️ Troubleshooting

Issue: Models downloading slowly

Solution: Pre-download models or use smaller variants (see Configuration)

Issue: High memory usage

Solution: Set environment variables for smaller models

Issue: RSS feeds not loading

Solution: Check internet connection; app uses cached data as fallback

Issue: Slow performance

Solution: Reduce n_items slider, increase refresh_sec, disable auto-refresh

See Full Documentation for more troubleshooting tips.

🗺️ Roadmap

📝 License

MIT License — See LICENSE file for details.

🙏 Credits

Built with:

Streamlit — UI framework
HuggingFace Transformers — AI reasoning
VADER — Sentiment analysis
FAISS — Vector search
Altair/Plotly — Visualizations
PyVis/NetworkX — Network graphs
Google News — RSS feeds

📞 Support

For issues or questions:

Check Full Documentation
Review Test Suite output
Inspect browser console (F12)
Check Streamlit logs in terminal

⚡ Nexora Intelligence Engine — Decision-Grade Intelligence, Powered by AI

Last Updated: 2025-11-07

Name		Name	Last commit message	Last commit date
.devcontainer		.devcontainer
assets		assets
engine		engine
lib		lib
models/models--sentence-transformers--all-MiniLM-L6-v2		models/models--sentence-transformers--all-MiniLM-L6-v2
.gitignore		.gitignore
FIXES_AND_IMPROVEMENTS.md		FIXES_AND_IMPROVEMENTS.md
NEXORA_GUIDE.md		NEXORA_GUIDE.md
QUICK_START.md		QUICK_START.md
QUICK_START.txt		QUICK_START.txt
QUICK_VERIFICATION.md		QUICK_VERIFICATION.md
Readme.md		Readme.md
SETUP_AND_FEATURES.md		SETUP_AND_FEATURES.md
TRANSFORMATION_SUMMARY.md		TRANSFORMATION_SUMMARY.md
UPGRADE_SUMMARY.md		UPGRADE_SUMMARY.md
YC_READY_SUMMARY.md		YC_READY_SUMMARY.md
app.py		app.py
app.py.backup		app.py.backup
requirements.txt		requirements.txt
streamlit_app.py		streamlit_app.py
test_core.py		test_core.py

Latest commit

History

Folders and files

Repository files navigation