Abstract of " Improved Scientific Analysis through Domain Driven Visualization and Support for Analytic Deliberation " Improved Scientific Analysis through Domain Driven Visualization and Support for Analytic Deliberation

of “Improved Scientific Analysis through Domain Driven Visualization and Support for Analytic Deliberation” by Radu Jianu, Ph.D., Brown University, May 2012. This dissertation introduces and evaluates novel visualization methods that enable researchers to derive and test hypotheses from available scientific data faster and more accurately than before. Following the traditional visualization approach, we introduce novel ways of visualizing and interacting with scientific data that support and accelerate researchers’ data analysis workflows. Following the visual analytics path, which advocates for supporting the reasoning process itself, we quantify the degree to which interface design elements can be used to unobtrusively guide researchers towards applying verified and established analysis techniques in their research. We first present novel visualization methods that were developed in response to analytic needs indentified through collaborative efforts in three concrete application areas. In neuroscience we enable faster interaction with diffusion tensor imaging (DTI) datasets by creating planar representations of the inherently 3D data. In proteomics we facilitate the visual collation of experimental data and existing protein interaction information and accelerate the discovery process by uncovering and supporting elements of the proteomic analysis workflow. In genomics we increase the accessibility of analyzable visualizations of microarray data and eliminate the overhead of creating visualizations and learning new systems by implementing and evaluating a novel data distribution method. Finally, we use the concepts of persuasive technology and “choice architecture” which state that a user of a system can be unobtrusively guided towards behavioral patterns that are more efficient, in terms of self-assumed goals, by slight alterations in the system interface. We provide quantitative experimental support for the hypothesis that we can use subtle changes in the interfaces of visual analysis systems to influence users’ analytic behavior and thus unobtrusively guide them towards improved analytic strategies. We posit that this approach may facilitate the use of visual analytics expertise to correct biases and heuristics documented in the cognitive science community. Improved Scientific Analysis through Domain Driven Visualization and Support for Analytic Deliberation by Radu Jianu B. S., Polytechnic University of Timisoara, 2005 Sc. M. Brown University, 2007 A dissertation submitted in partial fulfillment of the requirements for the Degree of Doctor of Philosophy in the Department of Computer Science at Brown University Providence, Rhode Island May 2012 c © Copyright 2012 by Radu Jianu This dissertation by Radu Jianu is accepted in its present form by the Department of Computer Science as satisfying the dissertation requirement for the degree of Doctor of Philosophy. Date David H. Laidlaw, Director Recommended to the Graduate Council Date Odest Chadwicke Jenkins, Reader Date Ben J. Raphael, Reader Approved by the Graduate Council