Show simple item record

dc.contributor.advisorBalazinska, Magdalena
dc.contributor.authorMorton, Kristi
dc.date.accessioned2016-03-11T22:38:53Z
dc.date.available2016-03-11T22:38:53Z
dc.date.submitted2015-12
dc.identifier.otherMorton_washington_0250E_15285.pdf
dc.identifier.urihttp://hdl.handle.net/1773/35165
dc.descriptionThesis (Ph.D.)--University of Washington, 2015-12
dc.description.abstractData has become more widely available to the public for consumption, for example, through the Web and the recent “Open Data” movement. An emerging cohort of users, called Data Enthusiasts, want to analyze this data, but have limited technical or data science expertise. In response to these trends, online visual analytics systems have emerged as a popular tool for data analysis and sharing. Current visual analytics systems such as Tableau and Many Eyes enable this user cohort to be able to perform sophisticated data analysis visually at interactive speeds and without any programming. Together, these two systems have been used by tens of thousands of authors to create hundreds of thousands of views, yet we know very little about how these systems are being used. The first challenge we address in this thesis, thus, is: how are popular visual analytics systems such as Tableau and Many Eyes being used for data analysis? To the best of our knowledge, this is the first study of its kind, and presents important details about the use of online, visual analytics systems. Visual analytics systems provide basic support for data integration. A simple approach for interactive data integration in Tableau was implemented in that tool in the context of this dissertation. Visual analytics, systems, however, do not currently assist users with de- tecting or resolving potential data quality problems including the well-known deduplication problem. Recent approaches for deduplication focus on cleaning entire datasets and com- monly require hundreds to thousands of user labels. In this thesis, we address the challenge of deduplication in the context of visual data analytics with an approach that produces significantly cleaner views for small labeling budgets than state-of-the-art alternatives. The key idea behind the approach is to consider the impact that individual tuples have on a visualization and to monitor how the view changes during cleaning.
dc.format.mimetypeapplication/pdf
dc.language.isoen_US
dc.subjectData cleaning; Data integration; Tableau; Visual Data Analytics
dc.subject.otherComputer science
dc.subject.othercomputer science and engineering
dc.titleInteractive Data Integration and Entity Resolution for Exploratory Visual Data Analytics
dc.typeThesis
dc.embargo.termsOpen Access


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record