Hybrid AI Approach for Classifying Web Content in Malaysian Public Universities

In the digital age, university websites are pivotal platforms for institutional communication, academic resource dissemination, and stakeholder engagement. This study presents a large-scale analysis of the official websites of 20 Malaysian public universities, encompassing 7.3 million web pages. Employing a hybrid methodology that integrates machine learning (XGBoost) and web-crawling techniques, we classified website content into nine categories: admission, e-learning, endowment, department, repository, events, news, research, and directory. Our findings reveal that directory pages overwhelmingly dominate the content landscape, indicating a sector-wide emphasis on administrative transparency and accessibility. Research and e-learning content are substantial in some institutions but are not the leading categories overall. The study also identifies underrepresented areas such as endowment, events, and news, highlighting opportunities for improved digital engagement. These content distribution patterns reflect the diverse strategic orientations of Malaysian universities and have implications for student recruitment, research visibility, and global rankings. The proposed framework offers actionable insights for optimizing university web strategies and serves as a replicable model for digital transformation in higher education.