hagosg/TigSen

Tigrigna sentiment Analysis Dataset # TigSen Tigrigna sentiment Analysis Dataset. This repository contains Python 3 code for performing Sentiment Analysis on Tigrigna social media texts using various deep learning and transformer-based models. The goal of this project is to classify the sentiments of posts or comments written in the Tigrigna language, one of the widely spoken languages in Eritrea and Northern Ethiopia. 📂 Repository Structure bash Copy Edit Tigrigna-Social-Media-Texts-Sentiment-Analysis/ │ ├── README.md # Project overview and usage instructions ├── TigSent_CNN/ # CNN-based sentiment classification ├── TigSent_LSTM/ # LSTM-based sentiment classification ├── TigSent_XLM_R/ # XLM-Roberta-based model (Transformer) ├── TigSent_mBERT/ # Multilingual BERT-based model ├── TigSent_LLaMA/ # Multilingual Cross transfer model ├── TigSen.xlsx # Labeled dataset of Tigrigna texts 🔍 Project Objectives - Fine-tune state-of-the-art NLP models to classify Tigrigna text sentiment as: - Positive 😃 - Negative 😠 - Neutral 😐 - Explore and compare the performance of: - Convolutional Neural Networks (CNN) - Long Short-Term Memory networks (LSTM) - XLM-RoBERTa transformer - Multilingual BERT (mBERT) - Multilingual LLaMA 📊 Dataset The dataset used is: - Tigrigna_Social_Media_DataSet.xlsx: A manually labeled dataset of Tigrigna sentences collected from social media platforms. - Each entry contains: - Tigrigna sentence - Sentiment label (Positive, Negative, Neutral) 🧠 Models Overview 🧱 TigSent_CNN A convolutional neural network model built using Keras/TensorFlow for sentence classification. 🔁 TigSent_LSTM An LSTM-based architecture to capture sequential dependencies in Tigrigna sentences. 🌍 TigSent_XLM_R Fine-tuned XLM-Roberta model, which supports many languages including Tigrigna. 🧠 TigSent_mBERT Uses Multilingual BERT, pre-trained on 100+ languages including Tigr …