Facets of Time - Making the Most of Time's Structure in Interactive Visualization
Time-oriented data are ubiquitous in many real-world problems from areas like business, medicine, or ecology. Problems involving time-oriented data need special attention because time data are different from other kinds of data. Time has an inherent structure, most prominently seen in the calendar aspect of time being composed of smaller granularities, like years and seasons. As these granularities influence natural and social aspects found again in the data, explicitly harnessing the structures of time in data analysis methods can greatly improve the amount of information gained. To develop innovations in how to deal with time-oriented data in Visual Analytics, the project DisCo has been pursued in the context of a certain set of users who handle application scenarios of time-oriented data analysis in different sectors. Those scenarios often involve ill-defined problems, and the structure of time is one of the few tools they have to solve those tasks. The novel visualizations are evaluated regarding the question whether they can fulfill these users’ needs. The structure of time is already a central focus of research in Temporal Databases. In that context, the concept of granularities was first laid out. Granularities can model several aspects of time’s structure—ordinal vs. discrete scales, linear vs. cyclic time, and above all calendars. We build upon this approved concept, but also add support for different points of view. Introducing the structure of time into the modelling of the visualizaton process is done by expanding established process-oriented frameworks. The goal in this thesis is to include the structure of time in interactive visualizations. We envision a broad range of any possible users and tasks that have a connection to this area. Many of the users and tasks we had access to stem from the research project DisCo, though. Still, those are only to be seen as examplary characteristics of the user and task space of Information Visualization in general. To gain a better understanding of the state of the art, but also to have a better way of describing our novel approaches, we include a detailed analysis how the visual exploration process is performed. The analysis consistes of a process-oriented framework as well as a taxonomy for visualizations that make use of the structure of time. The taxonomy is based on the several levels of the framework. Then, the taxonomy is applied to various state-of-the-art visualizations that make use of the structure of time. Based on the state of the art and its analysis from a theoretical as well as a usercentered point of view, new interactive visualization methods that are specifically designed to harness the structure of time are presented. These include methods to make use of the calendar aspect as well as showing different points of view. Much research regarding different points of view is focused on different sources, like different persons, who provide different values for the same point in time. Contrary to that, we research different views (usually from one source) that differ over their own timeline (for example, a person changing her mind). Further visualizations contained in this thesis have the task to disclose patterns from Data Mining methods to users, applying the structure of time for a better understanding. To confirm the usability, utility, and appropriateness of our results, we have performed two user studies and acquired a number of expert assessments. Further evaluation steps as well as the application of the current results are presented in our chapter of future work.
|Year of Publication||
Institute of Software Technology and Interactive Systems
Vienna University of Technology, Vienna