motazsaad/arabic-sentiment-analysis ? reverse-engineered prompt

Build me a small Python project for sentiment analysis on Arabic tweets, based on a TSV dataset of positive and negative tweets. I want the full flow from cleaning and preparing the text, including handling emojis and filtering mixed tweets, to training and testing a few simple models with NLTK and scikit learn.

It should be able to load the tweet files, turn the raw data into something usable, and then run a sentiment classifier that can predict whether a tweet is positive or negative. Please include clear notebook style examples or scripts for exploring the data, preprocessing Arabic text, and comparing a couple of basic approaches.

If you need to check current library docs online, go ahead. Keep it simple and practical, with code I can run in Python 3 and easy instructions for installing what’s needed.