Optimization of Higher Education Management Policies through Big Data-Based Sentiment Analysis

Abstract

This research utilizes big data and sentiment analysis to refine and optimize management policies in higher education sectors. It involves a comprehensive analysis of 52,181 comments gathered from social media across 16 provinces in Indonesia, focusing on stakeholders' perceptions regarding current educational policies. This study employs advanced analytical methods, including Naive Bayes, Support Vector Machine (SVM), and Logistic Regression, to process sentiments extracted from diverse digital interactions. The findings uncover a significant predominance of negative sentiment, accounting for 54% of the total comments. This indicates widespread dissatisfaction with the current educational frameworks. Positive sentiments account for 32.4%, and neutral                                    sentiments make up 13.6%, suggesting areas where educational policies are received favorably, alongside points of contention. Among the analytical models employed, Logistic Regression achieved the highest accuracy at 99.81%, followed by SVM at 99.79% and Naive Bayes at 94.36%. This performance underscores the capability of data analysis technologies not only to manage but effectively interpret vast volumes of unstructured data, thus providing actionable insights. The results from this study highlight the critical need for adapting educational policies that align more closely with stakeholder expectations and regional educational dynamics, thereby improving the quality and reception of educational governance.