Exploring multi-dimensional datasets can be cumbersome if data analysts have little knowledge about the data. Various dimension relation inspection tools and dimension exploration tools have been proposed for efficient data examining and understanding. However, the needed workload varies largely with respect to data complexity and user expertise, which can only be reduced with rich background knowledge over the data. In this paper we address the workload challenge with a data structuring and exploration scheme that affords dimension relation detection and that serves as the background knowledge for further investigation. We contribute a novel data structuring scheme that leverages an information-theoretic view structuring algorithm to uncover information-aware relations among different data views, and thereby discloses redundancy and other relation patterns among dimensions. The integrated system, DimScanner, empowers analysts with rich user controls and assistance widgets to interactively detect the relations of multi-dimensional data.