harisushehu/Hausa-deep-sentiment-analysis

# Sentiment Analysis in Low-Resource Languages: A Case Study on Hausa Texts ## Overview This repository contains the code for the paper "Harisu Abdullahi Shehu, Kaloma Usman Majikumna, Aminu Bashir Suleiman, Stephen Luka, Md. Haidar Sharif, Rabie A. Ramadan, and Huseyin Kusetogullari, Unveiling Sentiments: A Deep Dive into Sentiment Analysis for Low-Resource Languages – A Case Study on Hausa Texts", which is currently under review. ## Description ## Datasets (SentiHausa) The SentiHausa datasets used in this research can be accessed in the following folders: - Translated Words: 'sentence_polarity' - Stop Words: 'stopwords' - Sentences: 'opinion_lexicon' ### Models Pre-trained models can be accessed in 'models' folder. ### Bag-of-Words Approach - **BoW.py:** Investigates the sentiment of Hausa using the bag-of-words approach. ### Stemming Method - **BoW_stemmed.py:** Investigates sentiment analysis of Hausa texts with three categories (positive, negative, and neutral) using a stemming method proposed in our paper. - **BoW_stemmed_binary.py:** Investigates sentiment analysis of Hausa texts with two categories (positive and negative) using the same proposed stemming method. ### Deep Learning Models - **CNN.py:** Implements sentiment analysis using Convolutional Neural Networks (CNN). - **HAN.py:** Implements sentiment analysis using Hierarchical Attention Network (HAN). - **RNN.py:** Implements sentiment analysis using Recurrent Neural Networks (RNN). - ### Comparison with Other Studies - **comparison_other_studies.py:** Implements various machine learning algorithms, including Random Forest, Logistic Regression, K-nearest neighbors, Support Vector Machines, and Naive Bayes. These algorithms have been proposed in other studies and are utilized in this file to compare their performance with the methods employed in the current study. ### Statistical Testing - **ANOVA.py:** Conducts a statistical test to investigate the effectiveness of the proposed methods us …