Understanding relationships and commonalities between digital contents based on metadata is a difficult user task that requires sophisticated presentation forms. In this paper, we describe an advanced graph visualization that supports users with these activities. It reduces several problems of common graph visualizations and provides a specific chain arrangement of nodes that facilitates visual tracking of relationships. We present a concrete implementation for the exploration of relationships between images based on shared tags. An evaluation with a comparative user study shows good performance results on several dimensions. We therefore conclude that the ChainGraph approach can be considered a serious alternative to common graph visualizations in situations where relationships and commonalities between contents are of interest. After a discussion of the limitations, we finally point to some application scenarios and future enhancements.