Twitter Sentiment Analysis on Blended Learning in the Philippines using Hybrid Algorithm
John Paul Q. Tomas, Anna Georgina Lorenzo, Carlo Mangali, Louise Kathrene Roque · 2024
The recent pandemic, COVID-19, has affected different sectors in the Philippines, including the academic sector. DepEd and CHED were forced to implement hybrid learning to adapt to the current situation. This study aims to analyze the sentiments of Filipinos using Twitter sentiment analysis using Naive Bayes, Support Vector Machine, and NB-SVM. The steps through the process were Data Gathering, Data Cleaning, Data Annotation, Data-Preprocessing, Modeling, Model Testing, and Model Evaluation, accordingly. As for the results, the NB-SVM algorithm obtained the highest accuracy rate with 87.08%, while Naive Bayes and Support Vector Machine acquired 83.75% and 86.83%, respectively. Moreover, some words have a significant effect on the positivity and negativity inclination of each algorithm. Some words like learn, blend, and student were the highest three mutual positive words for all algorithms, while talaga, class, and school for negative inclinations. The researchers recommend further studies to explore the recent algorithms and to expand the period of data gathering.