Grant: Discovery Grants Program - Individual

Textual data is at the forefront of information management problems. Despite the improvements in the ways computers manipulate language data, our most common interfaces provide insufficient support, presenting information as long sequences of text for sequential reading. Information visualization has been successfully employed to provide interactive tools for summarizing and analyzing many types of data. However, existing visualization techniques do not make use of available computational methods for text analysis, relying instead of simple word counts. The lack of easy-to-use methods to access, explore, comprehend, use, share, and annotate large volumes of text data is likely one of the reasons for growing feelings of information overload. In order to solve this problem, we need to create new interfaces which present information in ways that facilitate reading, searching, analyzing, and annotating ever larger volumes of text data. Specifically, the proposed research will develop new techniques which combine linguistically-sophisticated algorithms with custom-designed interactive visual representations grounded in human perceptual capabilities. We will contribute new tools for analyzing large-scale online and streaming text datasets, from blogs to the proceedings of Parliament. The research will move beyond text databases, incorporating non-textual data such as illustrations from digital books, statistics from scientific experiments, and data related to text such as author names and dates. Combining both textual and non-textual data, we will present contributions which leverage complementary information in multiple types of data to present powerful new visual analysis methods for large mixed datasets. We expect up to 9 students, including undergraduate, master's, and doctoral to receive research training through participating in this research program.