Department of Computer Science

Aidan Slingsby

My Research

Dr Aidan Slingsby
Lecturer in Visual and Analytic Computing
Room: A304H
Department of Computer Science
School of Informatics
City University
London EC1V OHB
tel: +44 20 7040 0180
fax: +44 20 7040 8584

(Geo)visualisation for exploratory visual data analysis

I am interested in interactive visual techniques for exploring large datasets to allow spatial and temporal data to be viewed, filtered and inspected. I build software prototypes for demonstrate and test techniques for generating knowledge from data, usually based on case studies. I work with a variety of different types of data and user. Most of my work mades use of Processing - a Java-based design environment - which allow ideas to be prototyped quickly. I have also done work with APIs for creating mashups with an emphasis of exploratory data analysis.

Recent work has been on the use of appropriately ordered hierarchical displays for visualising multivariate data with spatial and temporal components. This work has resulted in a conceptual model for this encapsulated by the HiVE language and an implementation called HiDE - free software that implements HiVE, facilitates data exploration using hierarchical displays and allows these to be shared via blogs or Twitter.

Some examples of my work

Please see my news page and publication list.

Spatial Data Representation

I am interested in data models and strategies for representing space for different purposes.

My knowledge of this has been particularly useful for my recent work on building interactive and visual data exploration prototypes, as these have to handle large quantities of data efficiently (so that they are highly interactive) and generate (simple) statistical summaries in response to user interaction.

In my MSc project, I developed a model for computing flow paths and sub-basins on landscapes. The model was a triangular irregular network (TIN) implemented as an object-oriented data structure which contained basin- and flow-delineation routines. This enabled elements of the TIN to calculate their contribution to flow paths and sub-basins, allowing whole basins to be delineated in a piecemeal fashion. The use of the irregular TIN structure was to avoid the appearance of flow directional artefacts often associated with raster implementations. TIN structures also present some problems for flow routing. Where channel flow is along concave edges of TIN facets, the connectivity of the channel network is dependent on the triangulation configuration. My MSc project addressed some of these, and developed basin-delineation algorithms.

In my PhD project, I worked with the British Ordnance Survey to develop a data model and some specifications for 3D digitial mapping in urban area, incorporating building interiors. The model address geometrical, semantic and topological aspects of the built-environment including that of pedestrian access.

I recently participated in a research network led by the Technical University of Delft, in which we comparing different approaches to 3D modelling.

Spatial Uncertainty and its Visual Representaion

As part of the Willis Network, I am working on spatial uncertainty. In the context of data and models, uncertainty encompasses all that is unknown about the modelling process (known unknowns and unknown unknowns), including the uncertainties surrounding the data, the operation of models and the model outputs. Uncertainty is inherient in any model and the challenge to decision makers is to understand and therefore manage the uncertainty in order to make better decisions. Part of the understanding of uncertainty is the spatial variation of uncertainty and its correlation with other associated factors.

Where models output publish an assessment of the uncertainty surrounding the result, analysts' lack of understanding about uncertainties often leads to a reduction in confidence of the model outputs compared to those without published uncertainty, even though the former provides information valuable for decision-making. Appropriate visual representations of uncertainty may be assist in this process.