Hierarchical Aggregate Structure by Inductive Aggregation for Interactive Data Visualization

Data visualization is a method to facilitate the process of knowledge discovery and decision making. Effective and practical visual analytic systems have to support real-time and smooth interaction. Traditional data processing techniques and systems are inadequate when it comes to large datasets. The powerful features of relational database engines can be adapted to facilitate the visual representation of very large datasets.

In this thesis, we propose techniques to provide a tightly coupled system with a database engine back-end to support a data visualization front-end. We employ data reduction techniques along with other methods to form a hierarchical data structure that we call the inductive-aggregate pyramid to provide multiple representations of data. We also propose a generalized form of inductive-aggregate pyramids that we call cubed pyramids to provide a richer representation of high-dimensional data. We develop and employ techniques to build and query efficiently these structures to support interactive data visualization.