In the past few years, a growing amount of e-commerce information has been published online either as Linked Open Data or embedded as Microdata or RDFa markup inside HTML pages. Unfortunately, the usage of such data for product search and comparison is hampered by the products and services being themselves specific and heterogenous with regard to their relevant characteristics, and by the search process that involves learning about the option space. In this paper, we present an adaptive faceted search interface over product offers in RDF. Our search interface is directly based on the popularity of schema elements in the data and does not rely on a rigid conceptual schema with hard-wired product features, thereby being suitable for arbitrary product domains and product evolution. Further it supports learning during the search process. As a proof of concept of our work, we provide two use cases, namely one with product offers from an automobile database, and a second one with real product data collected from the Web.