Penerapan Metode Content Based Filtering pada Rekomendasi Menu di UMKM Me Love

Abstract

The culinary industry faces challenges in helping customers find menu items that meet their specific needs on digital platforms, as existing search features tend to be generic. Me Love, a spicy noodle shop offering a diverse range of options regarding flavor preferences and price points, currently lacks a mechanism for automatically recommending products to customers. This study developed a web-based menu recommendation system using the Content-Based Filtering (CBF) method and the Sørensen–Dice Coefficient algorithm to analyze product similarities based on attributes such as category, dish type, and spiciness level. The dataset comprised 50 active menu items. Functional testing via Black-box testing demonstrated that all key features of the admin dashboard and the recommendation filters operated as designed. Meanwhile, performance testing using a confusion matrix yielded a precision of 84% and an accuracy of 94.8%, indicating the system's ability to provide recommendations aligned with the product characteristics sought by users. This system is expected to enhance product search efficiency and the overall user experience within Me Love's digital catalog.