One of the goals of a recommender system is to minimize the cognitive load on the user and hence we cannot expect the users to give extensive feedback. The lesser the feedback, the lesser we know about the preferences of the user to make useful recommendations. This work aims to make the most use of user-provided feedback with the help of the hidden evidence in the casebase. The evidence for each product is acquired based on the relation among the products in the domain. The effectiveness of our approach is demonstrated through evaluation on three product domains. © 2019, Springer Nature Switzerland AG.