| Previous DGG based visualizations have not been completely successful. The most recent attempt, known as GenSpace, has three shortcomings. First, GenSpace is not dynamic. For a given set of DGGs, GenSpace is unable to dynamically generate both the 2D visualizations of the DGGs and the 3D visualization of the resulting generalization space graph (GSG). Secondly, GenSpace produces duplicate nodes when constructing a GSG visualization based on parallel path DGGs. Finally, GenSpace is not capable of producing a GSG visualization for merge path DGGs. This thesis addresses these specific research problems.; We address these issues as follows: First, in order to dynamically visualize an arbitrary set of DGGs, we have developed a set of visual criteria, to which a DGG visualization must adhere. Then, given this set of criteria, we extend the breadth first and depth first graph search algorithms to assign a 2D position to each node in the DGG. Once each DGG has a physical representation, a GSG visualization is constructed. To address the challenges surrounding the visualization of GSGs, we introduce the technique of rotational positioning. This technique allows for the construction of GSG visualizations using DGGs in their original form, regardless of whether the DGGs are single path, parallel path and merged path. (Abstract shortened by UMI.)... |
