Doctor's Theses (authored and supervised):

D. Ceneda:
"Guidance-Enriched Visual Analytics";
Supervisor, Reviewer: S. Miksch, C. Tominski, C. Collins, T. Dwyer; Institute of Visual Computing & Human-Centered Technology, 2020; oral examination: 2020-10-23.

English abstract:
As we continue producing data, the need to extract useful information from it has become essential. The implications of pursuing successful data analysis are before our eyes. We use it, although often unconsciously, when we compare products or check the availability of rooms for our next vacation. Data analysis is even more important for successful business, as it provides the foundation for solid decision making. Information visualization and visual analytics (VA) are two prolific branches of data analysis exploiting external means-namely, visualizations-to support the execution of analytical tasks. However, despite the vast adoption of such techniques, the issue with visual data analysis is that gaining insights is often quite a challenging process. Issues may arise at any phase of the analysis: Typical challenges include preparing data for the analysis or choosing appropriate analytical models and visual means for a given task. Users may encounter difficulties when exploring data and interpreting findings in light of domain knowledge, when organizing findings consistently into insights, when proving working hypotheses, and when generating new knowledge. The consequence is that, when issues arise, users are typically overwhelmed, the analysis stalls, and the efficacy of VA approaches is reduced; in addition, the quality of the resulting results is impaired.

For these reasons, it is important to support users with guidance. The aim of guidance is to foster an effective use of analysis tools and help users overcome any possible issue that might occur during the analysis. Historically, supporting users has been one of the very main goals of visual data analysis. However, how to reach this goal has never been researched in a systematic way. In this thesis, building on previous research and utilizing a user-centered methodology, we describe a thorough characterization of guidance approaches. Particular emphasis is given to guidance in VA. The main contributions of this thesis are: (1) we characterize guidance and summarize the main aspects of the process of guiding, also illustrating what constitutes guidance in the first place; (2) we report the effects of providing different types of guidance to users performing analytical tasks with different levels of expertise, shedding light on how guidance influences the way the analysis is conducted and how users react to it; and (3) we outline how guidance can be integrated with a step-by-step procedure in VA approaches-that is, directly at the point of designing VA tools. Although many open challenges still lie ahead, the results described in this thesis represent an initial demonstration of the value of guidance in VA and its positive effects on users while preparing a solid foundation for the development of inspiring guidance-enriched VA prototypes.

guidance, visual analytics, information visualization, visual data analysis

Electronic version of the publication:

Created from the Publication Database of the Vienna University of Technology.