Amdemichael/amharic-ecommerce-extractor

# Amharic E-commerce Data Extractor A comprehensive Named Entity Recognition (NER) system for extracting business entities from Ethiopian Telegram e-commerce channels. This project builds a smart FinTech engine that identifies the best vendor candidates for micro-lending by analyzing product names, prices, locations, and engagement metrics. ## 🎯 Project Overview **Business Need**: EthioMart aims to become the primary hub for all Telegram-based e-commerce activities in Ethiopia. This project develops an NER system to extract key business entities from unstructured Amharic text and provides vendor analytics for micro-lending decisions. **Key Objectives**: - ✅ Develop repeatable data ingestion workflow from Telegram channels - ✅ Fine-tune transformer models for Amharic NER (Product, Price, Location entities) - ✅ Compare multiple models and select the best performer - ✅ Implement model interpretability using SHAP and LIME - ✅ Create vendor analytics engine with lending scorecard ## 🚀 Quick Start ### 1. Setup Environment ```bash # Clone the repository git clone github.com cd amharic-ecommerce-extractor # Create virtual environment python -m venv .venv source .venv/bin/activate # On Windows: .venv\Scripts\activate # Install dependencies pip install -r requirements.txt ``` ### 2. Configure Telegram API Edit `config.yaml` with your Telegram API credentials: ```yaml telegram: api_id: YOUR_API_ID api_hash: YOUR_API_HASH phone: YOUR_PHONE_NUMBER channels: - '@ZemenExpress' - '@nevacomputer' - '@aradabrand2' - '@ethio_brand_collection' - '@modernshoppingcenter' ``` ### 3. Run Complete Pipeline ```bash # Run all tasks python src/main.py --task all # Or run individual tasks python src/main.py --task ingestion # Data collection python src/main.py --task training # Model training python src/main.py --task interpret # Model interpretability python src/main.py --task analytics # Vendor analytics ``` ## 📁 Projec …