Logo Lanfrica

bezawitsebsibe/Amharic_Ecommerce_Extractor

Domaine:

natural language processing

Type de record:

dataset
Créateur:
bez
Hôte:
# Amharic_Ecommerce_Extractor # Amharic E-commerce Data Extractor - Task 1 ## Goal Scrape and preprocess messages from Ethiopian e-commerce Telegram channels to build a clean dataset for Amharic NER model training. ## Completed - ✅ Scraped messages from 5 Telegram channels - ✅ Collected metadata: channel, timestamp, views, sender ID - ✅ Cleaned Amharic text and tokenized - ✅ Saved raw and tokenized messages to CSV - ✅ Downloaded images for product posts - ✅ Saved token preview sample ## Folder Structure