Information visualization is essential for understanding connections and patterns within large datasets, such as movies in the Netflix database, TV programs or social networks on Facebook.
However, traditional approaches such as pie charts, bar graphs and scatter-plots often don't show underlying patterns and relationships in the data.
My colleagues and I overcome these problems by using maps: aesthetically appealing visualizations that portray relations among abstract concepts. A map representation is familiar and intuitive; most people use maps; and well-drawn maps can provide hours of enjoyable exploration.
We automate the process: It begins with the data and ends with a computer-generated drawing of a map, rather than a graph. Our maps reveal some surprising insights into how different aspects of the data relate to one another.
Our approach has many different applications. In collaboration with AT&T Labs-Research, we have used our maps to visualize scientific collaboration networks, Netflix movies and books on Amazon. Working with UA nutritional scientists, we are identifying and mapping barriers to healthy eating and physical activity.
We are also, in collaboration with network engineers, using these map-based visualization techniques to better understand security threats to computer networks.
About the scientist
Stephen Kobourov is a professor of computer science at the University of Arizona. His research interests include information visualization and human-computer interaction. He has been at the UA since 2000. He spent a year at the University of Botswana as a Fulbright Scholar and a year at Tübingen University in Germany as a Humboldt Fellow.
• Stephen Kobourov's webpage: cs.arizona.edu/~kobourov
• Stephen Kobourov's mapping projects: cs.arizona.edu/~kobourov/PROJECTS/maps.html
• UA department of computer science: www.cs.arizona.edu