Skip to main content

Computer Science Seminar - November 2, 2016

Title: Exploratory Visual Text Analysis in the Scientific Literature Domain

Speaker: Florian Heimerl, University of Stuttgart, Germany

 

Abstract: Interactive visual text exploration and analysis approaches have experienced an impressive upsurge in the past decade. Such text methods are typically designed to support information search and retrieval scenarios. They assume that users carry out analyses with previous knowledge about what type of information they are looking for, and facilitate fast access to it. Open exploration, in contrast, that starts with no clearly defined objectives, and with little knowledge about the data set at hand, is widely unsupported. It requires detailed, user-driven analysis loops to allow users to get an understanding of its contents and gradually develop analysis interests and goals. This talk will present three approaches to support such exploration scenarios. The first one is an interaction technique that enables the exploration or document spatializations at arbitrary levels of granularity. The second project explores an approach to support human users with data foraging and filtering for information needs that manifest during exploration. The third approach has been specifically designed for the exploration of scientific literature data sets. It visually embeds the contents of scientific documents within their metadata contexts, facilitating joint exploration. Moreover, the talk will briefly introduce a data set of VIS publication, which is an open resource for scientific literature data from the VIS community.

 

Biography: Florian Heimerl is a doctoral candidate at the Institute for Visualization and Interactive Systems at the University of Stuttgart, Germany. He received his Diplom degree in computational linguistics from the Natural Language Processing Institute at University of Stuttgart in 2011. His research interests are interactive visualization and visual analytics of text data, with a particular focus of application on scientific literature collections.

 

University of Ontario Institute of Technology logo