Projects

Below is a summary of my (formal) projects, both ongoing and completed.

Measures and Algorithms for Effective Information Visualization using Visual Complexity and Trade-offs

VIDI project, funded by NWO

Proposal summary: Data is often visualized, allowing analysts to understand the data and make informed decisions. This requires good algorithms to create such visualizations automatically, algorithms that arrange all visual elements optimally such that the visualization is as legible as possible. However, visual complexity of the result is not properly accounted for: instead, simple indicators are used that do not relate well with visual complexity. This research will integrate visual complexity with algorithms for visualization, such that the complexity can be controlled explicitly, and facilitate a structural trade-off between different quality aspects. This allows portraying the underlying in the best possible way.

Analysis and Vizualization of Hetrogeneous Spatio-temporal Data

Commit2Data project with HERE, Fugro, Utrecht University and NDW, co-funded by NWO

See also: Commit2Data project page

Proposal summary: Movements of vehicles, people and animals are nowadays increasingly measured and stored, thanks to technologies such as GPS. An important example is digital traces (tracks) of the routes taken by cars. Not only the route points, but also the times of measurement are stored, so that speed along the tracks can be determined anywhere.

Analysis of such data is important for solving traffic issues, including preventing accidents and traffic jams. This is extremely relevant for road safety, the economy and quality of life in general.

In this project, researchers are working with three leading parties that collect and process traffic information (HERE, Fugro and the NDW - National Database Road Traffic Data), to jointly solve various issues, by modelling heterogeneous spatio-temporal data and calculating algorithmically in a generic way. These issues are categorised into three themes.

Real-time visualisation of flow data in their 3D context, such as 3D models of a city. This allows, for example, to relate speed to close proximity of vegetation or other factors that obstruct the field of view.
Detecting patterns related to traffic situations such as intersections and roundabouts for individual vehicles or clusters of vehicles. With this kind of pattern recognition, computers can actively help to look for, for example, dangerous situations or changes in driver behaviour.
Improving data quality using other data sources. This requires a geometric model for quality.

Algorithmic Geo-visualization: From theory to practice

Netherlands eScience project

Proposal summary: Visual representations, often in the form of maps, are one of the most effective ways for humans to interact with large data sets; they aid with complex cognitive tasks such as discovery and decision making. Information visualization plays a key role in exploring, analyzing and communicating large quantities of data.

Time-varying and dynamic data sets (for example, stock prices, traffic status, or weather) remain a challenge for most visualization algorithms. A primary requirement is stability: small changes in the data should lead to small changes in the visualization. Without stability, there is no cohesion between two visualizations showing similar data, which makes them difficult to interpret.

Visual analysis tools should be easily accessible to allow domain scientists across all areas, specialists and even the general public to explore, analyze and communicate data. However, many cutting-edge geo-visualization techniques are not available in an easy-to-use form; they exist at best as research proof-of-concept implementations. Furthermore, techniques for stable visualizations of dynamic data are still mostly lacking.

This project has two goals: 1. Develop two eScience tools (an online platform and a code library) to move advanced information-visualization and mapping techniques from theoretic concepts to practical tools, thereby increasing their impact through reusability 2. Significantly extend the state-of-the-art by developing stable geo-visualizations that can handle large quantities of time-varying data

Algorithmic Approaches to Spatially-Informed Information Visualization

Marie-Curie fellowship at giCentre (City University of London), funded by the ERC's Horizon 2020 programme