Information visualizations utilizing multiple coordinated views allow users to rapidly explore complex data spaces and discover complex relationships. Most multiple-view visualizations are static with regard to the views that they provide and the coordinations that they support. Despite significant progress in the field of Information Visualization and development of novel interaction techniques, user interfaces for exploring data have lacked flexibility. As a result, the vast quantities of information rapidly being collected in databases are underutilized and opportunities for advancement of knowledge are lost. This research proposes the central concept of visualization schemas based on the Snap-Together Visualization (Snap) model, analogous to the successful database concept of data schemas, which will enable dynamic specification of information visualizations for any given database without programming. Relational databases provide significant flexibility to organize, store, and manipulate an infinite variety of complex data collections. This flexibility is enabled by the concept of relational data schemas, which allow data owners to easily design custom databases according to their unique needs. We bring the same level of flexibility to visualization development through visualization schemas. Visualization Schemas is a conceptual model, user interface, and software architecture while Fusion is the implemented system that enable users to rapidly and dynamically construct personalized multi-view visualization workspaces by coordinating visualizations in ways unforeseen by the original developers.