The research area 'Visual Interactive Analysis' deals with the research and development of visualization techniques for the interactive exploration and analysis of large real-world datasets. Our work focuses on two directions of research:

• The SimVis project combines techniques from the fields of Scientific Visualization and Information Visualization for interactively analyzing time-dependent flow data.
• Purely Information Visualization-related work comprises projects with partners from the financial sector, designing and implementing a library for handling the internal data management, and extending the InfoVis capabilities of existing SimVis views.

In this page, we offer a list of current topics for student work. However, if you have own proposals for related topics, you are also welcome to get in contact with us.

• Interactive visualization and analysis of simulation data
• Combining scientific and information visualization
• Large, multi-dimensional and time-dependent data
• Linked views
• Direct Interaction by brushing data
• Further information: www.simvis.at

Integrating data derivation for datasets with moving geometry

Data derivation is an important feature of the SimVis system: It allows to calculate new data attributes by performing computations on existing ones (e.g. normalization, derivation). The challenge is to find meaningful ways for extending this approach to non-steady geometry.

This topic is suitable for a computer science project (Praktikum) or a master thesis (Diplomarbeit).

Contact: Helmut Doleisch (doleisch@vrvis.at)

SimVis extensions to handle streaming data

The current version of SimVis allows only analysis of data, which has to be completly available at the beginning of an analysis session. However in some applications, for example when working with real-time measured data (e.g. weather radar), or when binding tha analysis to a running simulation process, changeable, non-static data is generated. Streaming data (or data which changes at some specific points in time during an investigation) poses difficult questions, for example in terms of reapplying previous selections, changing scales for different data variables, validating and invalidating different data values (e.g. of different timesteps), and many more.

This topic is suitable for a computer science project (Praktikum) or a master thesis (Diplomarbeit).

Contact: Helmut Doleisch (doleisch@vrvis.at)

Improving the presentation backend of SimVis

The current version of the presentation view uses shaded glyph-based volume rendering to visualize features in the original geometry of the datasets. Many extensions are possible, among them adding interactive integration-based flow visualization techniques (like streamlines) and improved cutting plane visualizations.

This topic is suitable for a computer science project (Praktikum) or a master thesis (Diplomarbeit).

Contact: Helmut Doleisch (doleisch@vrvis.at)

Missing data visualization in existing SimVis views

Datasets with missing data values pose special challenges for visualization, starting with the question of where to place the visual representation of certain data items (e.g., a dot in a scatterplot) when the value of a data attribute (which the visualization depends on) is missing, or how to allow brushing of these entries. The goal is to extend existing views so that missing data is treated in an intelligent way.

This topic is suitable for a computer science project (Praktikum) or a master thesis (Diplomarbeit).

Contact: Helmut Doleisch (doleisch@vrvis.at) or Harald Piringer (piringer@vrvis.at)

Categorical data treatment in SimVis

As of now SimVis provides only very limited means to deal with non-numerical, especially with categorical data. Views like scatterplots, histograms, etc., become especially useful when the values of the data attributes spread out to the available visualization space. With categorical data, we often have the special case that a data attribute can only have one out of very few possible values, e.g., "yes", "no", "don't know", and "not specified". We aim at an integrated solution which allows to deal with numerical and non-numerical data at the same time.

This topic is suitable for a master thesis (Diplomarbeit).

Contact: Helmut Doleisch (doleisch@vrvis.at) or Harald Piringer (piringer@vrvis.at)

Comparison of simulation results

Currently, SimVis is capable of dealing with only one dataset at a time. However, some tasks require a comparison of several datasets (e.g. several results obtained with different parameters). It is an interesting and relevant matter of research how this feature can be integrated both in technical terms - handling several large datasets - as well as considering the semantics of a visualization.

This topic is suitable for a master thesis (Diplomarbeit).

Contact: Helmut Doleisch (doleisch@vrvis.at)

• Interactive visualization and analysis of abstract data
  • High-dimensional
  • Categorical
  • Large
  • Relational
• Multiple linked views
• Economical data (e.g. stock market data)

Extending the combined scatterplot

The combined 2D/3D scatterplot is an InfoVis view within the SimVis system that has been designed to visualize data as occurring in the context of flow-simulation data. However, some minor extensions - especially to the 2D visualization like alterable point sizes, density to transparency mapping, etc. - would suffice to make it an appropriate tool also for more general purpose data as needed for InfoVis applications.

This topic is suitable for a computer science project (Praktikum).

Contact: Harald Piringer, (piringer@vrvis.at)

Comparison of existing InfoVis solutions

There are various solutions for InfoVis available both in form of commercial products and as freeware. It is crucial for further developments to get an overview what is available and how well different kinds of data can be analyzed using the respective solutions - among other questions. Thus the goal of this work is to provide a state-of-the-art in InfoVis with special emphasis on available products.

This topic is suitable for a master thesis (Diplomarbeit).

Contact: Harald Piringer, (piringer@vrvis.at)

Focus and Context between views

Although a powerful concept, the technique of multiple linked views shows drawbacks when too many views are open at a time and the user runs out of space on screen and loses overview. This research has two goals: First, we want to develop a concept for arranging multiple views on screen in a way that reflects the needs of the user's work. Second, we want to find some kind of level-of-detail approach for GUI-elements that could be used to efficiently guide the user during the session.

This topic is suitable for a master thesis (Diplomarbeit).

Contact: Harald Piringer, (piringer@vrvis.at) or Helmut Doleisch (doleisch@vrvis.at)