A web-based skincare product recommendation using filtering

Abstract

On a large website, finding necessary information can be challenging and time consuming. To solve this problem, recommendation systems have emerged parallel to the development of the web and are now commonly used in several types of products. The recommendation is a strategy that makes suggestions based on the needs of the consumer to discover new and pertinent items for them by filtering through both non-personalized and personalized information or the user's preferences from a large amount of data. In this paper, the web-based skincare product recommendation is created by applying three effective approaches, Content-based Filtering, Collaborative Filtering, and Non-personalized filtering method, each offering a certain context for a recommendation. Skincare product customers often expect positive outcomes but may face challenges when their chosen products lead to adverse skin conditions, highlighting the importance of finding suitable options for their individual needs. The system described in this paper utilizes Collaborative Filtering to make recommendations based on the user's preferences, interests, or observed behavior regarding the item. Additionally, it incorporates Content-Based Filtering to suggest similar products based on ingredient similarity. It also provides a filtering method to make recommendations by brands, categories, and user characteristics. The implemented system successfully achieved its initial requirements by enabling clear communication between users and the system, providing recommendations for similar products, suggesting suitable products based on users’ skin characteristics, brands, and categories, and offering personalized suggestions based on their past behaviors. The application provides several helpful features, top 5, 10 and all suitable skincare products, but more improvements are still needed especially hybrid approaches. In future work, this system may be possible to increase suggestion accuracy by investigating hybrid systems that include both Content-Based Filtering and Collaborative Filtering methods. It is possible to create a recommendation system that is more thorough and individualized by combining the advantages of the two approaches.