Comparaison de la lisibilité des graphes en représentation noeuds-liens et matricielle
Résumé
This article describes a taxonomy of tasks on graphs and a controlled experiment for assessing the readability of two graph representations: matrices and node-link diagrams. The experiment only concerned a subset of the described tasks but provides important insights on the use of representations depending on the graph sizes and densities. It shows that, for graphs larger than 20 nodes or sufficiently dense, most tasks are performed more efficiently on the matrix-based representation. Only some path related tasks score better on node-link diagrams.