Pubster is an interactive tool for visualizing the relationships within a set of publication references. This project was done in collaboration with Amanda Alvarez for a visualization course at Berkeley. The system displays the set of citations as an undirected graph with each node representing an article and each edge encoding a shared attribute (author, keyword, etc). The software provides dynamic queries which allow the user to interactively select the attributes to use for structuring the graph. Search fields provide the functionality to query the data set for specific entries, and the user can retrieve details on specific articles by simply clicking on the nodes in the graph.
We implemented Pubster in Java using the Prefuse visualization toolkit written by Jeff Heer.
Here is a screenshot from the Pubster application:
For this example, we have loaded citations for the 20 most popular downloads from the Journal of Vision. In this image, we have chosen to structure the graph according to the keyword and year fields from the citation records. This means that an edge exists between two nodes if the two articles share a common keyword, or if they were published in the same year. We have also selected a specific node (Watson 2005) to show the details from this paper.