fuaSmart/amharic_ecommerce_data_extractor

# Amharic E-commerce Data Extractor ## Project Overview This project aims to build a robust **Amharic E-commerce Data Extractor**. The core objective is to transform unstructured and multimodal data (text and images) from various Ethiopian Telegram e-commerce channels into structured, machine-readable information. This extracted data will then be leveraged to populate EthioMart's centralized database, ultimately aiding in identifying promising vendors for micro-lending by providing insights into their activity and product offerings. Key focus areas include: * **Data Ingestion:** Programmatically collecting data from Telegram channels. * **Named Entity Recognition (NER):** Fine-tuning LLMs to extract Product Names, Prices, and Locations from Amharic text. * **Vendor Analytics:** Developing a scorecard to assess vendor potential for micro-lending based on extracted data and Telegram metadata. ## 📁 Project Structure This repository follows a standard data science project structure for clarity and maintainability: * `data/`: Stores raw (`raw/`) and processed (`processed/`) datasets. * `notebooks/`: Contains Jupyter notebooks for exploratory data analysis (EDA), prototyping, and interactive development. * `scripts/`: Houses standalone Python scripts for automated tasks (e.g., Telegram data scraping). * `src/`: Contains core Python modules for reusable functions and project logic. * `models/`: Stores trained machine learning models. * `reports/`: For generated reports and figures (`figures/`). * `tests/`: Unit and integration tests for code quality. * `config/`: Configuration files for project parameters. * `utils/`: General utility functions. * `.github/workflows/`: GitHub Actions for CI/CD pipeline. * `.vscode/`: VS Code specific settings. ## ⚙️ Setup and Installation To set up and run this project locally: 1. **Clone the repository:** ```bash git clone github.com cd amharic_ecommerce_data_extractor ``` …