A New Visual Language for Incremental Generation of Visual Representations

As the user base of visualization solutions expands, it is now even more critical to help end-users visualize their data quickly and effectively. Current automated visualization techniques use only a small set of basic guidelines mostly based on perceptual principles. Yet, visualization design principles and guidelines, as discussed in literature, cover a broad spectrum of factors that guide effective visualization design. In our previous research, we priorly analyzed existing principles and guidelines and examined factors that influence visualization design and found several limitations that the current visual principles include. To overcome these limitations, we propose a semi-automatic visualization approach that uses a declarative language for specifying visual representations and interactions with data. In this approach, visualization design principles and guidelines are expressed as rules that can encode human perception, information organization, data characteristics and semantics, and display characteristics, based on our examination of visualization literature and practice. By applying rules iteratively, a dataset is incrementally transformed into an interactive visualization via the application of data and visual operators. Intermediate representations of the specification, along with applicable rules, can be made shown to users to allow them choose the one that suits them best, which they can further adjust with minimal effort. Rules are easily extensible to customize and conform to domain-specific standards and practices. As a generative and language-based technique our approach offers flexibility not only in terms of design rules applied but also in the variety of visualization produced.