SIMS 247: Information Visualization

Group Project: Bibliographic Information Visualization and Analysis (BIVA)

Team: Chitra Madhwacharyula, Colleen Whitney, and Lulu Guo

Related work

Since we were doing an exploratory study, we iterated through several ideas before selecting a general approach. Our original idea was to adapt a tree map visualization, which would have provided an excellent macro-level view of the collection. One major issue with the tree map, however, was that they are not suitable for comparative views, as it is difficult to look at two tree maps and understand quickly what has changed.

We needed a visualization that could provide the macro-level view and facilitate comparisons, as well as providing for detailed item views. Fortunately, we were introduced to a visualization technique called PaperLens, developed by the University of Maryland and Microsoft Research.

PapleLens was developed to reveal trends, connections, and activity throughout a conference community. It could tightly couple different views across papers, authors, and references. There are three things we particularly liked about PaperLens. First, it had an overview of the collection along defined axes located in the left-hand side of the screen to let users capture a general pattern of the data in the collection given the criteria specified by the two axes. Secondly, PaperLens was capable of zooming in and seeing patterns within a subset in the right hand side of the screen. Lastly, it had the ability to highlight an item and see its details as well as additional patterns within the collection. (Fig.1)

Fig. 1. Paperlens

PaperLens provided a very effective way of exploring a large dataset while allowing user to drill down into more detailed information by using brushing and linking techniques. We found that this approach was especially suitable for our project since we were dealing with a large bibliographic collection and we needed an interface that gave users a good starting point in the left overview panel as well as the ability to zoom in on particular items of interest in the collection. Our project shared with PaperLens the goal of revealing trends and patterns among items in a collection, so although there were differences in the number and types of variables and the size of our dataset, it was an effective model for developing our prototypes.