David Furcy, Andrew Jungwirth, Thomas Naps
BlockTree is an information visualization tool that helps students explore nuances of heuristic search algorithms, such as A* and WA*. The goal of such exploration is to achieve a level of understanding that is deeper than merely being able to trace the algorithm on an artificially small domain. Instead, BlockTree presents a single execution snapshot of a heuristic search tree encompassing up to hundreds of thousands of nodes and offers a robust GUI that encourages viewing such a tree from a variety of perspectives. By experimenting with the visualizations generated by the problem domains already incorporated into BlockTree, students are exposed to the same large-scale issues that AI practitioners face on a regular basis. Beyond these built-in problem domains, BlockTree offers an extensible object-oriented framework that can be used as the basis for programming assignments. As students implement additional domains in BlockTree, their solutions are automatically visualized in the same fashion as the built-in domains.
Subjects: 15.7 Search
Submitted: Feb 9, 2007