Skincare product recommendation system using rule-based algorithm

With the rapid development of the commercial skincare business, the diversity of skincare products is overwhelming, and consumers are often confused and face a higher risk of skin adverse reactions such as contact dermatitis due to the lack of dermatological knowledge. Modern methods such as machine learning and collaborative filtering are commonly used for product recommendation, however, they are “black box” models that lack transparency and might lead to dangerous and unpredictable results for sensitive skin types. This project tackles these difficulties by inventing Skinalyze, an automated safety-critical facial skincare product recommender system based on a deterministic rule-based (IF-THEN) inference engine. The system is developed using the Agile Software Development Life Cycle (SDLC) methodology and operates on a secure Three-Tier Architecture using a Python stack. The presentation tier is built on the Streamlit framework with a limited number of dropdown inputs to minimise user data entering errors. To improve the data privacy, user parameters are captured in the volatile session states and are wiped clean on the browser shutdown. These inputs are then passed to the logic tier, which compares them to a knowledge base associated with an embedded SQLite relational database that contains structured product formulas extracted from curated public datasets. The parameterised SQL queries were hard-filtered, which resulted in a flawless precision rate; algorithmic verification was performed using a 20-combination cross-matrix test map. User Acceptance Testing (UAT) under institutional ethical approval received high mean values for Ease of Use, Relevance of Results and Trust in Logic. Furthermore, a bivariate Pearson correlation analysis was conducted, revealing a significant positive linear link between a user’s impression of backend rule safety and their commercial intent to follow the suggestions. The results show that transparent explainable AI (XAI) frameworks can safely bridge the consumer knowledge gap and provide a scalable architecture that can be replicated in other safety-critical domains such as dietetics or pharmaceuticals.