The visualization of relational data by node-link diagrams quickly leads to a degradation of performance at some exploration tasks when the diagrams show visual clutter and overdraw. To address this challenge of large-data graph visualization, we introduce Graph Metric Views, a technique that enriches the visualization of traditional layout strategies for node-link diagrams by additionally allowing an analyst to interactively explore graph-specific metrics such as number of nodes, number of link crossings, link coverage, or degree of orthogonality. To this end, we support an analyst with additional histogram-like representations at the axes of the display space for graph-specific metrics. In this way, a cluttered and densely packed node-link diagram becomes more explorable even for dense graph regions: The user can use the distribution of metric values as an overview and then select regions of interest for further investigation and filtering.
@inproceedings{Panagiotidis2014GraphExplorationMultiple, author = {A. Panagiotidis, M. Burch, O. Deussen, D. Weiskopf, T. Ertl}, booktitle = {2014 18th International Conference on Information Visualisation}, doi = {10.1109/IV.2014.51}, issn = {1550-6037}, keywords = {data visualisation;graph theory;relational databases;degree of orthogonality;dense graph regions;display space;exploration tasks;graph exploration;graph metric views;graph-specific metrics;histogram-like representations;large-data graph visualization;layout strategies visualization;link coverage;link crossings;metric values;multiple linked metric views;node-link diagrams;performance degradation;relational data visualization;visual clutter;Clutter;Data visualization;Histograms;Joining processes;Layout;Measurement;Visualization;graphs;metrics;node-link diagrams}, month = {jul}, pages = {19--26}, title = {Graph Exploration by Multiple Linked Metric Views}, url = {http://graphics.uni-konstanz.de/publikationen/Panagiotidis2014GraphExplorationMultiple}, year = {2014} }