Machine Learning Approaches for Nepali News Categorization: Naive Bayes and Support Vector Machine
Keywords:
Feature Extraction, Naive Bayes, Nepali Corpus, Nepali News Classification, Support Vector MachineAbstract
Purpose: This study explores performance of Support Vector Machine (SVM) and Na¨ıve Bayes (NB) classification techniques for Nepali news classification.
Methods: To experiment the system, news were collected from different online social media news portals. We analyzed user interactions with news posts to identify patterns and preferences across different domains, specifically health and politics.
Results: Our study evaluates effectiveness of these classification models based on accuracy, precision, recall, and F1-score. Results indicate that while SVM generally provide better classification performance with 91.5% accuracy, Na¨ıve Bayes with an accuracy of 85.3% remains a competitive alternative due to its simplicity and efficiency.
Conclusion: Our research work applies SVM and Na¨ıve Bayes models to classify Nepali news enabling automated categorization of news articles into predefined categories.
