Democratizing Details-on-demand Data Visualizations at Scale

Abstract

Details-on-demand is a powerful interaction paradigm which features the use of simple mouse interactions such as pan and zoom to help the viewer navigate through a large data space. In the past years, we have witnessed an increasing amount of data visualization applications that embrace this paradigm to facilitate data exploration and analysis. Web maps are a clear example. However, due to the highly specialized nature of these applications as well as the lack of general scalable toolkits, building new details-on-demand data visualizations remains hard especially for large datasets. This thesis proposes new tools and systems to “democratize” details-on-demand-based data visualizations, i.e., to make it much easier to build such applications at scale. The main approach is to offer declarative data visualization languages for developers to author applications in small amounts of code, and work with a database backend to transparently handle the rendering and performance optimizations needed to enable fluid interactions on large datasets.