Improving Consumer Decisions through Preference Relaxation

In online shopping scenarios, it can be difficult for consumers to process large amounts of information and make a good purchase decision, particularly for multi-attributed products selection. Interactive decision aids that support preferential choice using information filtering are a potential solution to this problem. However, current methods that enable filtering based on initial customer preferences may eliminate potentially valuable alternatives early in the decision process, and possibly negatively impact decision quality. Drawing from recent developments in recommender systems and query rewriting, we propose a new method that minimizes the elimination of high quality alternatives early in the decisionmaking process. We discuss the model and the potential impact of our method based on an evaluation scenario with 56915 real car adverts. Finally, we give an overview of the implications of our study for practitioners and researchers.