PhD

dissertationManaging and understanding the large amounts of scientific data is undoubtedly one of the most difficult research challenges scientists are facing today. Data exploration through visualization is an effective means to understand and obtain insights from large collections of data. Although several visualization tools are available, including tools with sophisticated visual interfaces, they are out of reach for users who have little or no knowledge of visualization techniques and/or who do not have programming expertise. In addition, as interdisciplinary groups work together, the ability to generate a diversified collection of analyses for a broad audience in an ad-hoc manner is essential for supporting effective scientific data exploration. Science portals and visualization web sites have been used to simplify this task by aggregating data from different sources and by providing a set of predesigned analyses and visualizations. However, such portals are often built manually, and are not flexible enough to support the vast heterogeneity of data sources, analysis techniques, data products, and user communities that need to access this data. The goal of this dissertation is to provide a complete framework for streamlining the creation of customized visualization applications and to facilitate collaboration and sharing among scientists and visualization experts. The framework is composed of three main components: VisMashup, for creating customized visualization applications based on dataflow specifications; Provenance-Rich Publications, a component that allows users to create documents (web pages, presentations or pdf files) whose digital artifacts (e.g.,figures) include detailed provenance information (workflow and associated parameters) used to produce the artifact; and CrowdLabs, a system that adopts the model used by social Web sites and that combines a set of usable tools and a scalable infrastructure for providing a rich collaborative environment for scientists and that also takes into account the requirements of computational scientists, such as accessing high-performance computers and manipulating large amounts of data. To demonstrate how users can benefit from this framework, a series of use case scenarios is also included