A hybrid recommender system for an online store using a fuzzy expert system

Nowadays, various recommender systems are popular and their main aim is to recommend suitable content to the user based on various parameters. This article proposes a hybrid recommender system, Eshop recommender, which combines a recommender module composed of three subsystems (the subsystems use collaborative -filtering and content-based approaches) and a fuzzy expert system. It is an e-shopping recommender system for suggesting suitable products. The system works with different user preferences and their activity on the e -shop, and the resulting list of recommended products is created using the fuzzy expert system. The expert system works with several parameters -similarity level with already rated products, coefficient of purchased product, and an average rating of the product. Due to this, our proposed system achieves promising results based on standard metrics (Precision, Recall, F1-measure). The system achieves results above 90%. The system also achieves better results than traditional approaches. The main contribution is creating a comprehensive hybrid system in the area of product recommendation in an online store, which has been validated on a group of real users and compared with other traditional approaches and the recommendation module of another online store.

Automated summary

This article proposes a hybrid recommender system that combines collaborative filtering, content-based approaches, and a fuzzy expert system. By analyzing user preferences and activity, and using the fuzzy expert system to create a recommended product list, this system achieves promising results based on standard metrics, even outperforming traditional approaches.