Description

Networks are a powerful and common representation for real world phenomena such as social relations, biological interactions, brain functionality or traffic flows. Nodes in the network can represent individual entities such as persons, proteins, brain regions or places while links between the nodes can model a plethora of relationships: constant flows or exchanged entities, static or dynamic relations, instantaneous events, simultaneous relations, strong or weak relations, and so forth. Networks help us model and express a plethora of information about the real world, yet networks are one of the most complex data sets to understand.

This PhD topic sets out to investigate interactive and visual interfaces to better understand network data. Visualization allows to express complexity of the data in a different way than statistics and mathematical models. How noisy is the data? Where are the interesting parts? What can I expect from my data? How complete is my data? These questions in turn allow for the creation of hypothesis that can inform a subsequent in-depth analysis.

Besides this exploratory aspect of visualization, there are others such as supporting collaborative discussion of the data, monitoring changes in the data, and presenting findings to a larger audience.

The goal of this thesis is to design, evaluate, and apply cutting-edge novel visualization and interaction techniques to help network analysts, e.g., in biology, neuroscience and history, to understand network data. Visualizations can integrated into our visualization and research platform networkcube, and projects such as Vistorian. Both projects allow us to evaluate novel techniques with domain experts and improve through iterative design.

The scope of this thesis can be broad and is subject to the candidate's interests and interested collaborators. Possible angles for a thesis are:

• Dynamic Networks: Changes in relation over time, including changing edge weights, exchanged objects, propagation and flows, as well as changes in node attributes and general network characteristics [4,5,7,8].
• Geographic Networks: Positioning nodes and routing edges on geographic maps, such as social networks and their changes over time. [10,11]
• Storytelling with Networks: How to present scientific findings in networks to a larger audience such as lay people or the scientific community? How can visualization web-technology help in disseminating insights? [6]
• Annotation and Collaboration: How to support collaborative exploration of network data? How to support personal annotations and histories reflecting the exploration process and making it easier to the analyst to recap and continue his exploration after a longer break?
• Visualization of network measures: Measures exist to quantify various aspects of networks, such as density and degree distribution, as well as individual measures for nodes and links. While usually considered complementary to visualization, how can we use these measures to drive visualization design? [4]

Soft Requirements

The ideal candidate is highly interested in complex networks as well as the underlying phenomena. He or she potentially brings experience in related domains (digital humanities, biology, etc..) and information visualization (e.g., internships, practice or undergraduate research). Good coding skills are required for this thesis as well as interest in expressing and discussing ideas, creating and exploring design spaces, sketching, working with domain experts, as well as scientific evaluation, writing, and dissemination.

Preferably, visualizations should be coded using state-of-the art web technology (e.g., D3, WebGl, python) and made available to end users. Publications are meant to be submitted to top-tier conferences and journals in the field of Information Visualization (IEEE VIS, EuroVis, TVCG, etc.) and Human Computer Interaction (e.g. ACM CHI), and potentially others.