1. WebQuery: Searching and Visualizing the Web through Connectivity Rick Kazman Software Engineering Institute Pittsburgh, PA 15213 +1 412-268-1588 rkazman@sei.cmu.edu Jeromy Carriere Nortel Ottawa, ON Canada K1Y 4H7 +1 613-763-6305 jayc@nortel.ca

2. VISUALIZATION OF RESULTS Bullseye algorithm 3D graph 2D cone tree springs-and weights algorithm

3. Bullseye This visualization shows three concentric circles. –each containing nodes with equal connectivity –labelled with the connectivity value for that subset –the nodes with the highest connectivity are shown closest to the center

4. Figure 1: "Bullseye" view of "Rick Kazman" query.

5. 3D graph layout Figure 2(a) shows a sorted list view of the neighbor set. –All nodes with equal connectivity values are colored the same and grouped together in the list. Figure 2(b) shows a three-dimensional visualization in which nodes are laid out in planes according to their connectivity values. –nodes appearing in the original hit set are shown as tetrahedra –other nodes--those added as a result of WebQuery augmenting the query results--are shown as cubes

6. Figure 2: Visualization of "software engineering AND software architecture" query.

7. 2D cone tree Figure 3 shows a top view of a cone tree visualization of the resulting information space. The cone tree is constructed from the neighbor graph. Each artificial subset parent node (visualized here in green) is labelled by the connectivity value of its children. Nodes from the hit set that appear in the tree are colored according to their score. –red indicating highest score –blue indicating lowest score

8. 2D cone tree Nodes in the neighbor set that did not appear in the hit set from the original query are colored black. We not only visualize the connectivity of returned nodes.

9. Figure 3: 2D cone tree visualization of "library" query.

10. Springs-and weights graph layout algorithm The artificial parent nodes are shown in blue- green and labelled with the connectivity of their children, and the artificial root is explicitly labelled. This node is colored red, indicating that it has the highest connectivity value. the advantage that it tends to place highly connected nodes in the center of the viewing area. –nodes repel each other by default –connections cause nodes to attract each other

11. Figure 4: Graph visualization of "back pain" query.