The performance of recommender systems in online shopping: A user-centric study

• We evaluate four approaches to online product search in an extensive user study with real data.
• The approaches that we study exploit novel preference relaxation methods to recommend products.
• We compare the results of a user experiment with simulations, using two large sets of products.
• We show that a particular preference relaxation method improves the decisions of online shoppers.

This research investigates the effects of preference relaxation on decision-making performance of users in online preference-based product search contexts. We compare four recommender systems based on different preference relaxation methods in extensive user experiments with 111 subjects that use two real-world datasets: 1818 digital cameras and 45,278 used car advertisements gathered from popular e-commerce websites. Our results provide new insights into the positive impact of the Soft-Boundary Preference Relaxation methods on decision-making quality and effort. The paper extends previous studies on this topic and demonstrates that decision aids based on preference relaxation techniques can effectively enhance preference-based product search in online product catalogues and help alleviate common disadvantages of form-based filtering mechanisms.