Database updates using interactive Pan and Zoom visualizations

Abstract

As datasets continue to get larger, there is a great need for visualization systems that scale well while maintaining interactivity. Kyrix [1] is a system that helps developers create scalable pan and zoom visualizations. It combines visualization paradigms for good performance like data tiling and prefetching with a database backend that creates spatial indexes for faster querying. This thesis modifies Kyrix to support direct manipulation of datasets through a data visualization. We add bindings to the Kyrix specification that allow the developer to enable updates in a visualization. The user can then interact with the visualization and update the data in order to test a hypotheses, add new data, or fix a data error. We showcase this functionality by implementing three different types of Kyrix visualizations: an NBA timeline visualization, an election forecasting visualization, and a scatter plot visualization of NBA game scores. We report the update performance statistics for each of the demo visualizations and provide an evaluation of changes to the Kyrix specification language.