Back to Projects

Tunisian Sentiment Classification

A machine learning model to classify Tunisian dialect text as positive or negative.

Machine LearningNLPTunisian DialectHugging Face

Project Overview

This project, developed in collaboration with Khadija Kriaa and Yessine Bouattour, presents a sentiment analysis model trained specifically on the Tunisian dialect. It can classify a given text as Positive or Negative. The model is hosted on Hugging Face and is accessible via their API.

Try the Model

Enter some Tunisian text below (tunisian native language are supported, e.g., "chnahwelek") and click predict.

Try the Model

How it works

The model uses a fine-tuned transformer architecture (MARBERT) trained on a dataset of Tunisian social media comments. It handles the unique linguistic features of the dialect, including code-switching and specific vocabulary.