Paarweiser visueller Vergleich von gerichteten, azyklischen Graphen: Entwicklung von Gestaltungsrichtlinien basierend auf Mensch-Maschine Interaktion
Zusammenfassung der Projektergebnisse
Graphs have become an indispensable model for representing data in a multitude of domains, including biology, business, financing, and social network analysis. In many of these domains humans are repeatedly confronted with the need to visually compare node-link representations of graphs in order to identify their commonalities or differences. Yet, despite its importance little is known about how much visual differences affect users’ perception of graph similarity. As a result, more a systematic investigation addressing this issue is necessary. This was the goal of this project. We specifically concentrated on visual comparison of directed acyclic graphs. Within the project, we developed methodology for conducting visual comparison studies. We assessed the advantages and disadvantages of crowdsourcing and laboratory studies, developed tools for the generation and selection of study datasets, for conducting studies and for measuring perceived graph similarity. We conducted studies identifying factors, which influence human judgment of graph similarity for three cases: small unlabeled graphs, small labeled graphs and larger unlabeled graphs. Our results indicate that both graph-theoretic and visual factors influence the similarity judgment. On the graph-theoretic side, the number of levels, number of nodes and the labels of central nodes are important. On the visual side, graph shape and white space seem to play an important role. The review of guidelines for network visualization and the results of our studies served as a basis for conceptual work: review of visualization guidelines, the characterization of data and tasks in visual graph comparison and influences on mental models. Moreover, it opens new research questions: the development of novel visual comparison techniques that adhere to the identified human factors.
Projektbezogene Publikationen (Auswahl)
-
Crowdsourcing for information visualization: Promises and pitfalls. In Evaluation in the crowd. Crowdsourcing and human-centered experiments, pages 96–138. Springer, 2017
Rita Borgo, Bongshin Lee, Benjamin Bach, Sara Fabrikant, Radu Jianu, Andreas Kerren, Stephen Kobourov, Fintan McGee, Luana Micallef, Tatiana von Landesberger, et al
-
Investigating graph similarity perception: A preliminary study and methodological challenges. In 12th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications, VISIGRAPP 2017, pages 241–250. SCITEPRESS-Science and Technology Publications, Lda., 2017
Tatiana von Landesberger, Margit Pohl, Günter Wallner, Martin Distler, and Kathrin Ballweg
-
Visual data comparison. Habilitation, TU Darmstadt, Germany, 2017
Tatiana von Landesberger
-
Visual similarity perception of directed acyclic graphs: a study on influencing factors. In International Symposium on Graph Drawing and Network Visualization, pages 241–255. Springer, 2017
Kathrin Ballweg, Margit Pohl, Günter Wallner, and Tatiana von Landesberger
-
Insights by visual comparison: The state and challenges. IEEE computer graphics and applications, 38(3):140–148, 2018
Tatiana von Landesberger
-
Visual similarity perception of directed acyclic graphs: A study on influencing factors and similarity judgment strategies. J. Graph Algorithms Appl., 22(3):519–553, 2018
Kathrin Ballweg, Margit Pohl, Günter Wallner, and Tatiana von Landesberger
-
Perception of Differences in Directed Acyclic Graphs: Influence Factors & Cognitive Strategies. In Proceedings of the 31st European Conference on Cognitive Ergonomics (ECCE 2019). Association for Computing Machinery, New York, NY, USA, 57–64
Günter Wallner, Margit Pohl, Tatiana von Landesberger, and Kathrin Ballweg
