Diversity-enhanced recommendation interface and evaluation

The beyond accuracy user experience of using recommender system is drawing more and more attention. For example, the system interface has been shown to associate positively with overall levels of user satisfaction. However, little is known about how the interfaces can constitute the user experience and the social interactions. In this paper, I plan to propose a visual diversity-enhanced interface that supports the user to inspect and control the multi-relevance recommendations. The goal is to let the users explore the different relevance prospects of recommended items in parallel and to stress their diversity. Two preliminary user studies with real-life tasks were conducted to compare the visual interface to a standard ranked list interface. The users’ subjective evaluations show significant improvement in many metrics. I further show that the users explored a diverse set of recommended items while experiencing an increase in overall user satisfaction. A user-centered evaluation was used to reveal the mediating effects between the subjective and objective conceptual components. The future plans are discussed to extend the current findings.