Interactive Exploration of Complex Datasets Via the Effective Generation of Text and Graphics

North Carolina State University, Raleigh NC

Investigators

Abstract

This is accomplished through an interlocking collection of textual and graphical discourse engines, a plan recognition system, and an interaction manager that harnesses the power of these tools. Users begin by asking questions about the data stored in their dataset. The system answers using a combination of text and graphics. Text responses are built by a discourse engine, while graphical images are constructed using a perceptual visualization assistant. The collection of all possible responses is evaluated to select the most effective answer, be it text, graphics, or a combination of the two. A plan recognition system is used to analyze the users' queries and their reactions to the responses they receive. This allows the system to anticipate future queries, cache relevant statistics, and guide the discourse and visualization systems during the evaluation and selection of appropriate answers to each user query. Results from this project will include: (1) plan recognition and interaction plan construction performed by the system to identify and model current and future analyses conducted by the users; (2) presentations that are sensitive to both the current and anticipated future state of users' investigations; (3) assisted navigation techniques; (4) methods for evaluating the effectiveness of the use of text and/or graphics; and (5) perceptual visualization techniques. Results will be disseminated through journal and conference publications, online datasets with results and analysis, and online software demos of relevant research components. Although this project will study applications from the oceanography and public school domains, these results are relevant in any situation where interactive exploration of large, complex datasets is required. http://www.csc.ncsu.edu/faculty/healey/NSF-IDM-00