Spatial and Temporal Analysis and Visualization

Geo-visualization: Fostering Meaning, while Facing Imperfection and Big Data

Over the period, we have focused on enhancing the users’ experience with spatiotemporal information, providing them with highly interactive tools for exploration and knowledge discovery. We have particularly focused on the design and development of geovisualization techniques adapted to the rendering of geographic phenomena and geohistorical data, whether they are societal or environmental.

Among the challenges we have addressed, a first necessity is to consider
design approaches supporting the diversity and heterogeneity of data sources that describe events or phenomena to render. Historical data we had to manage in the ANR projects Biblindex and Eclats, in the project BDHI or in C. Saint-Marc’s PhD are characterized by various scales (both in space and time) but also imprecision and incompleteness. We have also addressed the visualization of contributive data collected from social networks (see C. Cavalière’s PhD) and of observations and measurements acquired from sensors in order to analyse recent territorial dynamics. All the approaches we have developed have in common to tackle these issues at a conceptual level to improve the identification, analysis and understanding of territorial dynamics, and to lead to the proposition of interactive visualisation models implemented in several prototypes.

Among the main results of this axis, J. Gautier’s, C. Chagnaud’s and A. Menin’s PhD have contributed to develop innovative ways of producing and interacting with visualizations. J. Gautier has proposed Graphist, a prototype allowing users to interactively explore a dataset and, choosing values for time, space and semiological parameters, to visually identify recurrence of events, spatiotemporal concentration of events or propagation process.

Beyond propositions related to the graphical restitution, this work is also an algorithmic contribution to the extraction of spatio-temporal patterns in a dataset which has been tested on very di↵erent case studies (hydrometeorology, accidentology and criminology).

C. Chagnaud’s PhD focuses on the development of statistical methods to analyze, explore and visualize geolinguistic data. He proposes various spatial interpolation methods to deal with qualitative data in order to create isogloss and visualize the spatial distribution of linguistic phenomena at various granularity levels. These methods are integrated in a cartographic environment whose aim is to considerably reduce the time linguistic experts spend on creating and customizing analysis maps by providing a set of automatic tools. The algorithms we have developped (for instance for calcutating the areal distribution of a phenomenon from qualitative ponctual data) can apply to any comparable datasets which are quite common in Digital Humanities.

A. Menin works on tangible interfaces aiming at multiplying the interactivity potential of geovisualisation tools. She has designed and developed Estime, a geovisualization prototype people can use for analysis purpose with a tablet as remote control, exploiting for instance a gyroscopic sensor to navigate in datasets. This prototype has been applied to a study on daily mobility. In this context, the tool has been used to co-visualize some (anonymized) data representing the activities and commute chains of a pool of individuals together with some aggregated data highlighting complementary points of view on territorial dynamics at different
geographic unit levels.

On this topic, S. Chardonnel (Lab. Pacte, Geographer, Steamer invited member) and L. Nedel (UFRGS, Brasil, researcher in HCI and scientific visualisation) are strongly involved. Another complementary direction we have investigated in these works refers to innovation in semiology, especially for graphically representing events that occurred at the same place through time, considering also their imprecision. This requires solutions to manage the superimposition of their footprints which has been explored in the environmental field, for flood risk or volcanic eruptions representation using both dynamic and static maps, and in the field of Digital Humanities, for the restitution in 3D – inspired by the Time-Geography – of biblical authors’ life trajectories (Early Christian Literature data, ANR project Biblindex).

Finally, we have investigated how to improve the assessment of our semiological propositions and prototypes. This led us to the definition of guidelines for a more efficient experimental protocol design in the field of geovisualisation.