Analyzing facts and opinions in Nepali subjective texts

Subjectivity Analysis is a relatively new field of research for the Nepali language. It offers a challenging area which has not been adequately studied till date systematically. Limited works that have been conducted in Nepali include works primarily on polarity detection [7]. In this work, we propose a Supervised Machine Learning based framework for analyzing facts and opinions for Nepali subjective texts. We train three different models using three Supervised Machine Learning Classifiers: (Logistic Regression, Multinomial Naïve Bayes, and Support Vector Machine) and conduct a comparative study based on the metrics: Accuracy, Precision, Recall and F-Measure. Our results show that the task of analyzing subjective sentences and making a distinction between facts and opinions can be conducted with reasonable accuracies close to 70%.