Personalized Skincare Recommendation System Based on Ontology and User Preferences

Personalized skincare product selection remains a complex but critically important challenge, as tailoring recommendations to individual skin profiles directly enhances treatment efficacy and fosters consumer trust. Traditional systems, such as content-based and collaborative-filtering, often fail to capture semantic interactions among skin types, concerns, and ingredients. To address these limitations, we propose an innovative ontology-based skincare recommendation system that integrates structured dermatological knowledge with semantic reasoning. Leveraging the Methontology framework, we developed an ontology composed of twelve core classes such as Product, Ingredient, Skin Type, and Skin Concern and more than twenty-five object properties to model interrelated concepts. The knowledge base was populated via web scraping from three prominent platforms (Sociolla, Beautyhaul, Skinsort), yielding over 3,800 products and 28,000 ingredients. We augmented this dataset with dermatological literature to ensure clinical validity. The architecture employs Apache Jena Fuseki and SPARQL for inference, with a React-Node.js web interface. Users input skin type, concerns, and sensitivities, which are translated into RDF triples and processed through semantic rules to generate personalized recommendations. An evaluation based on the Technology Acceptance Model (TAM) assessed Perceived Usefulness and Ease of Use. Ten diverse respondents rated the system with an average score of 4.5 out of 5 (SD=0.3) and endorsed the relevance of recommendations with a score of 4.8. Our findings demonstrate that semantic technologies can significantly enhance personalization and transparency in skincare solutions. This work lays a robust foundation for future innovations in beauty technology, clinical decision support, and consumer health platforms.