Corpus building and NLP analysis for Persian Telegram channel messages. Includes a notebook to parse and clean channel_messages.json, stopword normalization, word cloud with a silhouette mask, and CSV outputs (filtered_messages.csv, final_results.csv). Reproducible pipeline for EDA and basic modeling.
machine-learning data-mining deep-learning sentiment-analysis transformers pytorch topic-modeling bert persian-nlp news-classification text-preprocessing hazm parsbert telegram-data
-
Updated
Oct 14, 2025 - Jupyter Notebook