Ontology based Machine Learning Approach for Facial Skincare Products Recommendation

The need to preserve facial skin health and improve attractiveness has become more widespread in the modern world. There has been competition among skincare companies to research and create novel products. The dynamic skincare market creates a wide range of skincare products. Therefore, selecting the best skincare products suitable for a consumer's skin type and condition can be quite challenging. Skin impairments could worsen if products contain ingredients inappropriate for the user's skin type. This study proposes a unique system architecture for recommending facial skincare products that combine ontological and machine-learning benefits. The ML engine is developed for facial skin condition identification, which uses a CNN model to identify acne severity based on the user's facial image. An ontology was constructed using the Protégé ontology editor, which included hierarchical relationships between user profiles, skincare information, and skincare product information. The semantic similarity between these concepts was mapped by the Protégé tool for the skincare product recommendation engine. The system UI takes inputs for customization and recommendation of facial care products based on key factors such as the user's skin type, concerns, acne severity level, and allergy ingredients. Users can provide feedback and ratings for the products recommended. The developed system had an accuracy of $87.5 %$ based on a survey conducted with 24 participants who tested the system. Therefore, the proposed ontology-based ML approach is effective and accurate for facial skincare product recommendations.