Multilevel Compound Tree – Construction Visualization and Interaction

Abstract

Abstract. Several hierarchical clustering techniques have been proposed to visualize large graphs, but fewer solutions suggest a focus based approach. We propose a multilevel clustering technique that produces in linear time a contextual clustered view depending on a user-focus. We get a tree of clusters where each cluster -called meta-silhouette -is itself hierarchically clustered into an inclusion tree of silhouettes. Resulting Multilevel Silhouette Tree (MuSi-Tree) has a specific structure called multilevel compound tree. This work builds upon previous work on a compound tree structure called MO-Tree. The work presented in this paper is a major improvement over previous work by (1) defining multilevel compound tree as a more generic structure, (2) proposing original space-filling visualization techniques to display it, (3) defining relevant interaction model based on both focus changes and graph filtering techniques and (4) reporting from case studies in various fields: co-citation graphs, relateddocument graphs and social graphs.

Multiscale Visual Analysis of Lexical Networks

“…Graph hierarchies are also used in Focus-based multilevel clustering. In [6,7,8] several hierarchical clustering techniques are proposed, to visualize large graphs. These contributions are mainly concerned with accounting for a user focus in the construction of a multi-level structure.…”

“…These contributions are mainly concerned with accounting for a user focus in the construction of a multi-level structure. Sometimes this results in new multi-level structures such as for example MuSi-Tree (Multilevel Silhouette Tree) in [8]. Other approaches are based on zooming strategies that include level-of-details dependant of one or more foci [12].…”

TreePlus: Interactive Exploration of Networks with Enhanced Tree Layouts

“…It is mainly designed for graphs whose nodes are visual elements such as images. Boutin et al also used a radial layout to visualize a hierarchical clustered graph that is the result of multilevel clustering [7]. It is a graph of clusters, where each cluster itself is hierarchically clustered.…”