Sistem Rekomendasi Produk Makeup Berbasis Content-Based Filtering dengan TF-IDF dan Cosine Similarity

Abstract

The growing cosmetics industry offers a wide range of makeup products; however, consumers often face difficulties in selecting alternative products that align with their preferences. This study aims to develop a content-based filtering recommendation system to assist users in finding relevant products when their primary product is unavailable. The method includes data collection from Kaggle using the Luxxify Makeup dataset, data exploration, preprocessing, feature extraction using TF-IDF, and product similarity calculation using cosine similarity. The recommended products are those with the highest similarity based on category and product description. Evaluation was carried out using the Mean Average Precision (MAP) metric to assess the relevance of recommendations. The results show that the system successfully recommends five alternative products with very high accuracy (MAP = 1.00). This system contributes to providing a personalized and efficient product search solution and can be applied to e-commerce platforms or digital beauty services.