recommender system, e-shop recommender system, hybrid recommender system, content-based filtering, collaborative filtering, products

This article compares different approaches for recommending suitable products in the e-shop recommender system. The proposed recommender system consists of two main modules - a module for recommending suitable products based on viewed products and a module recommending suitable products based on the rated products. The first module uses a content-based filtering approach, and the TF-IDF algorithm is used for product recommendations. The second module uses a collaborative filtering approach, and the SVD algorithm is used for product recommendations. Furthermore, three approaches that combine the results of both modules are proposed. These approaches were experimentally verified on a group of 32 real users. The users tested the proposed recommender system and marked the relevant products from the list of recommended products. The results of the experimental verification are discussed.