The importance of effective customer assistance technologies is imperative in today's online marketplace where users are oftentimes overwhelmed by the product choices available to them. Relating their subjective preferences to the precise product descriptions poses an additional challenge, one which leads us to look at how research from two complimentary research communities (recommender systems and intelligent user interfaces) can be married to improve online recommender systems. In particular, we are interested in content-based recommendation domains that rely heavily on explicit feature-level feedback to narrow the number of relevant products for a user. A user's inability or unwillingness to provide detailed fine-grained information challenges applications in these domains and as such the way in which products are presented to the users and how these products are selected for presentation must adapt to suit this type of domain and user. Here we introduce the iCARE System, which provides an combination of product visualization techniques and additions to the current methods of user preference extraction to recommend suitable eyeglasses to individual users.