About HypoVis

HypoVis is a 3-years research project funded by the FWF Austrian Science Fund as project number P22883. We started our work in January 2011, and will continue till December 2013. Our goal is the development of new methodologies for Visual Analytics that are focused on time-oriented data and the structure of time.

The primary goal of Visual Analytics is the close intertwinedness of human reasoning and automated methods. In order to explain this, research in Visual Analytics has dealt with formulating a description for the Visual Analytics process. Our goal is to provide methods for making the transfer between all the states in the Visual Analytics process possible, with the focus of generating models from hypotheses. These models can, among other things, be used to validate the hypotheses.

As a basis for our work, we use a description of the Visual Analytics process we have developed based on work by Keim et al. [2008] as well as Bertini and Lalanne [2009]. To form hypotheses based on data, users have to apply interactive visual interfaces. Those are usually also used to help users understand models that result from automated methods like Data Mining. Our goal is to strengthen other ways of generating elements of the process, like the externalization of hypotheses and their validation.

Visual Analytics Process

HK: Take hypotheses from domain knowledge. VK: Visualize domain knowledge directly. MK: Take models from domain knowledge. VD: Visualize data. MD: Generate models from data. HV: Build hypotheses based on visualizations. VM: Visualize models. VH: Visualize hypotheses. MH: Validate hypotheses to form models. IH: Gain insights from hypotheses. IM: Gain insights from models. IV: Gain insights from visualizations.

Our goal is the design of a Visual Analytics system that uses the inherent structure contained in time-oriented data as a way to improve the integration of human reasoning. This structure can, for example, be seen in the calendar aspect of time being composed of smaller granularities, like years and seasons. In the following example from Aigner et al. [2007] the discrete time domain is shown with with multiple granularities. The smallest possible unit (chronon) is one day. Based on this, the granularities weeks and fortnights are defined.