ViMaL - The Visualization Mapping Language

Visualization is the discipline dealing with the depiction of data. The assignment of visual abstractions to data abstractions is referred to as the visualization mapping. Expressive visualization mappings proofed in many cases to be more effective than the ex-amination of raw data (i.e., large tables of numbers). Although visualization mapping is a cru-cial step in the visualization pipeline surprisingly few general approaches exist. We propose to develop the Visualization Mapping Language (ViMaL) that is capable of describing expressive visualization mappings and provides user interfaces suitable for non-experts in visualization. ViMaL will be a language that is used to specify visualization pipelines. We will provide language concepts for the specification of data abstraction processes, visual concretization processes and the required visualization mapping. With this approach the semantics of a specific domain are modeled by the domain experts while the semantics of the visualization domain are modeled by the visualization expert. The visualization mapping is described using domain semantics as well as visualization semantics. Unlike other general purpose visualization systems, ViMaL systems explicitly use the semantics of the visualization mapping process. It incorporates information and knowledge assisted methods in the visualization mapping pipeline. In the scope of this project we will define a novel Visualization Mapping Language and implement a toolbox that aids researchers from other areas to integrate the ViMaL concepts into their systems. The toolbox will provide basic ViMaL document setup and parsing as well as components for each step in the visualization pipeline. We will make use of fuzzy logic in our components for the abstraction of data, for the visual concretization, and for the visualization mapping. We believe that the formalization of the visualization pipeline that is achieved with the visualization mapping language will enable many researchers from other domains to benefit from existing visualization methods. Further, the introduction of domain and visualization semantics for the specification of visualization mappings enables meta-visualization approaches that provide insight into the visualization process itself. Laypersons will benefit from such meta-visualization systems that are able to illustrate the involved components.

Visualization is a tool used in many fields, where abstract data needs to be analyzed by domain experts through manual examination. By visualizing the data, they can be understood better and made useful for the domain experts. In order to visualize the data, it is essential to assign visual properties to the individual data points, so that the abstract data can be transformed to displayable images. This transformation is described by a set of rules called visualization mapping. Creating this description is a highly specialized task performed by computer scientists. However, the resulting images should serve as scientific tools for experts from various domains and their proper construction requires the expert knowledge from the given domain. The knowledge transfer between the domain experts and the computer scientists creating the visualization mappings is therefore necessary. This poses a burden on the workflow of all involved parties. The results of this project provide means for circumventing this problem by allowing the domain experts to describe the way how their data should be visualized using their own vocabulary. This removes the necessity of transferring their knowledge to computer scientists, which significantly simplifies the process of designing new visualizations for specific tasks or new types of data. The central element is the visualization mapping language, called ViSlang. This language gives the computer scientist the possibility to prepare the environment for domain scientists, which they can use to create their own visualizations. ViSlang allows the users to integrate languages specific for different domains in a single document describing the visualization mapping. This design effectively reduces the inter-domain knowledge transfer. Additionally to the language itself, we implemented various auxiliary tools for supporting domain experts in the specification of visualization mappings and the interaction with the visualizations. Since visualizations are often complex and depend on many parameters, we proposed several tools for easing the specification of the parameters for domain scientists, who might not be familiar with the precise internal workings of their software systems. Our goal was to allow the domain experts to interact with the parameters in relation to their domain knowledge, instead of the computer science knowledge. Additionally, we identified that an important task within the design and evaluation of visualization mappings is comparative visualization. This means to support the users in comparing outputs of sets of visualizations, created by different visualization mappings, or using different parameter settings. Within this project, we designed several tools which aid in efficient comparative visualization.