Caractérisation de visualisations et exploration interactive de grandes quantités de données multidimensionnelles

Abstract : A great number of activities use IS (Information Systems) relying on interactive graphic visualizations. As a vector for the transmission of information, these visualizations aim at optimizing the bandwidth between the user and the visual information. To create effective visualizations, designers use methods based on their expertise and user tests. These methods affect the design process, they are time consuming, and they do not rely on specific detailed theories, which could avoid, in advance, errors of conception. In order to better study these visualizations, we propose to characterize them, which means to systematically number and define their properties explaining all the visual phenomena that enable transmission of information. The thesis that we hold to, consists in saying that the characterization of a visualization is important for their conception, and that their exploitation in the light of these methods and tools enable the creation of better interactive systems. Thus, we have determined that a characterization is composed of two elements: its contents and the way in which its content take form. We will show that the DataFlow model, a model that usually permits the description of the construction of the visualization, does not accomplish a sufficiently complete characterization. We will propose other models enabling characterization of the dynamic aspects, emerging information (notably those issuing from so-called "ecological" design) and especially graphic accumulation. In using visualization issuing from the milieu of air transportation, we will show how the characterization enables the understanding of their similarities and differences, to remove the ambiguities from the communication between designers and to conceive of, for example, a new schematic visualization of air routes. In addition, we have adapted the process to transform a collection of data into a visualization in the form of an exploration software of large numbers of multidimensional data. This software relies notably on an exploration of visual configuration and on a collection of original interactions with powerful implementation. FromDaDy (From Data to Display) enables the configuration and manipulation, in real-time, of rich and original visualizations, such as those using graphical accumulation.
Document type :
Theses
Complete list of metadatas

Cited literature [95 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00610623
Contributor : Christophe Hurter <>
Submitted on : Friday, July 22, 2011 - 5:24:40 PM
Last modification on : Monday, April 29, 2019 - 5:05:02 PM
Long-term archiving on : Friday, November 4, 2011 - 3:15:22 PM

Identifiers

  • HAL Id : tel-00610623, version 1

Collections

Citation

Christophe Hurter. Caractérisation de visualisations et exploration interactive de grandes quantités de données multidimensionnelles. Interface homme-machine [cs.HC]. Université Paul Sabatier - Toulouse III, 2010. Français. ⟨tel-00610623⟩

Share

Metrics

Record views

599

Files downloads

1081