AkliluJo/Wolaita-political-sentiment-analysis

his repository implements a supervised machine learning-based sentiment analysis model for political news data in the Wolaita language # Sentiment Analysis for Political News Data in Wolaita Language Using Machine Learning This repository contains the code, resources, and documentation for the MSc thesis by Sintayehu Petros (Wolaita Sodo University, April 2025). The project develops a supervised machine learning model for analyzing sentiments (positive, negative, neutral) in Wolaita-language political news data, addressing challenges in low-resource languages. The best-performing model, Random Forest with TF-IDF and random oversampling, achieves 96.88% accuracy. The thesis PDF is included in the `docs/` folder for reference. Code examples are provided in `src/` (adapt as needed for your implementation). ## Abstract (Excerpt from thesis page xiii) Sentiment analysis, a pivotal tool in natural language processing (NLP), particularly in the political domain, has been extensively studied for high-resource languages but remains underexplored for low-resource languages like Wolaita. While some sentiment analysis research has been conducted in Wolaita, it has not specifically addressed political news data. This thesis aims to fill this gap by developing a robust sentiment analysis model tailored for Wolaita-language political news using supervised machine learning techniques. The study begins with the creation of a high-quality annotated dataset of 2,080 Wolaita political news sentences gathered from Wolaita Wogeta FM and Fana Broadcasting Corporation, Wolaita branch, curated with assistance from language experts and validated through inter-annotator agreement metrics. This dataset serves as a foundational resource for training and evaluating sentiment analysis models that consider the unique linguistic characteristics of Wolaita. The research systematically explores various machine learning models, including Logistic Regression, Support Vector Machines (SVM), Random Forest, and Multi-Layer Perceptron (MLP) Classifier, employing feature extraction techniques such as Bag of Words (BoW), Term Frequency- …