From Tree to Graph - Experiments with E-Spring Algorithm

Graph drawing and visualization represent structural information as diagrams of abstract graphs and networks. E-Spring Algorithm, derived from the popular spring embedder model, was proposed to eliminate node overlaps in the drawings of clustered directed acyclic graphs Gc. In this paper, we apply the E-Spring algorithm to general graphs by minimizing edge-node intersections. Initially, a tree structure is extracted from the original graph using the breadth-first search (BFS) algorithm. The extracted tree is then visualized without node overlaps using the E-Spring algorithm, and the remaining non-tree edges are appended to this visualization. A post-processing step that implements edge routing is performed on the obtained visualization to eliminate residual edge-node intersections. This method has been validated by visualizing eBay buyer-seller relationships and Graph Catalog benchmarking data.