Hybrid Recommender System utilizing Naive Bayes Classification and Weighted Content-Based Filtering
Year: 2026
Type: article
Abstract: Conventional tourism promotion often marginalizes lesser-known destinations by disproportionately favoring heavily commercialized landmarks. To address this inequity, this study designed and evaluated a dynamic destination recommender system tailored for the province of Antique, Philippines. The developed platform utilizes a hybrid approach, integrating a Naive Bayes (NB) classification algorithm with weighted content-based filtering (CBF) to match explicit user preferences against a comprehensive local database. Trained on a localized dataset of historical tourist visitation logs, the predictive model achieved high classification accuracy and an excellent F1-Score, outperforming traditional pure baseline models. The system was subsequently deployed and evaluated under the ISO/IEC 25010 quality framework by a group of respondents, yielding highly acceptable mean scores. The hybrid algorithm successfully democratized digital visibility, actively routing off-the-beaten-path niche destinations to highly receptive users. While Students and Tourism Officers reported flawless usability and functional suitability, Information Technology (IT) Experts assigned more conservative ratings regarding backend capacity and interoperability. The study concludes that the hybrid recommender successfully mitigates mainstream tourism bias, supports sustainable local tourism, and satisfies operational requirements.
Institution: University of Antique
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7220128882