报告人:    Jing Yang, The University of North Carolina at Charlotte

报告时间:  2012年5月6日星期日10:00am~11:00am

报告地点:  清华大学FIT 楼1区315房间

主办单位:  清华大学软件学院计算机图形学与辅助设计研究所

 

 Dr. Jing Yang is an associate professor in the Computer Science Department at the University of North Carolina at Charlotte. She received her PhD degree in computer science from Worcester Polytechnic Institute in 2005. She has been conducting research in the fields of visual analytics and information visualization. Her research interests include the visual reasoning process and large scale evolving text collections, graphs, and high dimensional data visual analytics. Her work has been published extensively in refereed journals and conferences. She was on the program committee for IEEE Information Visualization Symposium 2005 and 2006, IEEE Information Visualization Conference 2007-2011, IEEE Visual Analytics Science and Technology Conference 2011 and 2012, EuroVis 2010-2011, and IEEE Pacific Visualization Symposium 2011. She was on the organizing committee of IEEE Information Visualization conference 2007-2011. She is the lead guest editor of Tsinghua Science and Technology Special Issue on Visualization and Computer Graphics.

 

Abstract: Taking the true needs and goals of users into consideration can greatly simplify the visualizations and lead to effective and efficient visual analytics approaches. In this talk, two visual analytics systems developed following this guideline will be presented. The first system, EventRiver, allows users to interactively explore large temporally evolving document collections. It was designed, developed, and evaluated around the goal of supporting users in browsing, searching, tracking, and investigating real life events motivating the text generation. The second system, PIWI, visualizes large graphs without clutter. Motivated by the needs of analyzing real-world networks based on their community structure, PIWI closely integrates visualizations with automatic community detection. A set of uncluttered, intuitive visualizations and interactions based on communities are proposed to support tasks such as community-community relationship analysis, community-attribute relationship analysis, and scalable node selection according to complex structural feature and node attribute criteria.