Scientific knowledge embraces all the notions of the scientific subjects, such as mathematics, physics, chemistry and more. This knowledge requires to be visualized to various users and for different tasks. The purpose of this work is to improve the design of scientific knowledge visualization systems through the creation of User interface (UI) prototypes from the scientific knowledge to visualize. To achieve this purpose a methodology is proposed. The formal representation and visualization of this knowledge are the first research questions addressed by two reference models: Scientific Knowledge Model and Visual Model. A mapping language designed to declare abstract representations of the desired UI prototypes associates these two models. Algorithms to transform these abstract representations into concrete UI prototypes have been also developed. The methodology has been tested using various sources and the results, including the real use case of the CERN CMS High Level Trigger configuration management system, are presented.