Skip to content

README.md

Latest commit

 

History

History
120 lines (93 loc) · 3.48 KB

README.md

File metadata and controls

120 lines (93 loc) · 3.48 KB

LocalityAI: Hyperlocal Ad Strategy Engine

Streamlit Demo
Python APIs

AI-powered platform generating data-driven ad strategies by synthesizing:

  • 🌍 Geospatial Data (HERE Maps)
  • 📰 News/Events (NewsAPI, Ticketmaster)
  • 🚦 Traffic Patterns (HERE Traffic)
  • 💬 Social Sentiment (Reddit API + VADER)
  • 🤖 LLM Augmentation (Llama-3 via Together.ai)

🎥 Demo

LocalityAI Demo

Images

Summary Generated Image Generated

🔑 API Configuration

All API keys in this repo are placeholders. You must obtain your own:

  1. Required APIs (Free tiers available):

    # .env EXAMPLE
HERE_API_KEY=your_here_maps_key          # https://developer.here.com/
TICKETMASTER_API_KEY=your_ticketmaster_key # https://developer.ticketmaster.com/
NEWS_API_KEY=your_newsapi_key            # https://newsapi.org/
TOGETHER_API_KEY=your_together_ai_key    # https://together.ai/
REDDIT_CLIENT_ID=your_reddit_client_id   # https://www.reddit.com/prefs/apps
REDDIT_CLIENT_SECRET=your_reddit_secret

  2. Weather API Alternative (Used in code but requires registration):

    # get_weather_info.py uses WeatherAPI (https://www.weatherapi.com/)
WEATHER_API_KEY=your_weatherapi_key

🚀 Quick Start

git clone https://github.com/yourusername/LocalityAI.git
cd LocalityAI

# Install dependencies (check requirements.txt)
pip install -r requirements.txt

# Configure API keys
mv example.env .env  # Add your real keys here

# Launch Streamlit UI
streamlit run index.py

# Or run Flask API (separate terminal)
python flask_app.py

🧩 Code Structure

├── index.py                # Streamlit dashboard
├── flask_app.py            # Flask API endpoint
├── llm_response.py         # Llama-3 prompt engineering
├── geocode_location.py     # HERE Maps geocoding
├── get_weather_info.py     # WeatherAPI integration
├── get_news_info.py        # NewsAPI client
├── get_events_info.py      # Ticketmaster scraper
├── get_traffic_info.py     # HERE Traffic analysis
├── get_sentiment_info.py   # Reddit + VADER sentiment
└── requirements.txt        # Python dependencies

🌐 API Workflow

sequenceDiagram
    User->>Streamlit: Enter Location + Vertical
    Streamlit->>Flask: POST /get_data
    Flask->>HERE Maps: Geocode Location
    Flask->>NewsAPI: Fetch Headlines
    Flask->>Ticketmaster: Get Events
    Flask->>Reddit: Analyze Sentiment
    Flask->>Together.ai: Generate Strategy
    Together.ai-->>Flask: LLM Summary
    Flask-->>Streamlit: JSON Response
Loading

⚠️ Critical Notes

  1. Rate Limits

    • HERE Maps: 100k reqs/month (free)
    • NewsAPI: 500 reqs/day (free)
    • Reddit: 60 reqs/minute (OAuth2)

  2. Cost Monitoring

    # Together.ai: ~$0.20 per 1k tokens (Llama-3)
# HERE Maps: $0.50 per 1k geocodes after free tier

  3. Ethical Considerations

    • Reddit data anonymized via clean_text()
    • Location fuzzing (±100m) in get_sentiment_info.py

📜 License

Apache 2.0 - See LICENSE
Commercial use requires API subscriptions