Formal Comparison of Matrix- and List-Based Visualizations of Bipartite Graphs

Type: 
DA
Persons: 
1

Description

A bipartite graph is a special class of graphs, where the vertices can be partitioned into two independent sets. Data sets representing bipartite graphs can be found in many disciplines, ranging from biology, where nodes represent genes and conditions, document analysis, where nodes can represent different categories of named entitites, to social network analysis, where nodes can be institutions and projects.  

The most typical representations of bipartite graphs use lists or matrices. Yet, little is known which representation is more suitable for different tasks and data characteristics.

Tasks

The goal of this work is to design and perform a controlled experiment to assess the benefits and limitations of the two different visualization techniques. The following steps have to be conducted:

  • Literature research to investigate typical tasks during bipartite graph visualization (a starting point can be the task taxonomy for graph visualization by Lee et al., 2006).
  • Desiging the experiment.
  • Implementation of list- and matrix-based bipartite graph visualizations to generate stimuli for an experiment.
  • Conducting the experiment, for instance as crowd-based evaluation using Amazon Mechanical Turk.
  • Statistical evaluation and interpretation of the experiment results.

Requirements

  • Strong interest in human-computer interaction and visualization topics.
  • Prior experience with user studies is an advantage.
  • Knowledge of or strong will to learn statistics is a must.

Environment

The experiment stimuli can be generated using light-weight visualization tools, like d3. The experiment itself can be hosted through Amazon Mechanical Turk or a custom evaluation environment for a self-hosted experiment or lab experiment. The statistical evaluation can be conducted using any statistical software package, like R, Python, or SPSS, depending on the preference of the student.

Contact

For more information please contact Manuela Waldner (waldner@cg.tuwien.ac.at).