Improvise Background Architecture Examples Downloads
Example — DaVinci
icon
Click image for full-size version.
Background
As part of a project to understand sense making and decision making involving conflicting opinions in text documents, we looked at occurrences of positive and negative words and phrases in reviews of Dan Brown's “The Da Vinci Code”.
Interface
The main feature of this visualization is a variation on the basic two-sided arc diagram in which additional time series information is displayed between the positive and negative sides.
In the modified arc diagram, increasing time is labelled from left to right along the center axis. The top half of the diagram shows terms used in positive review, the bottom half terms used in negative ones. Arcs connect months in which common terms appear. Arc thickness represents the number of common terms. Bar thickness shows the number of terms for each month, considered individually. (Bar thickness tends to be thicker than arc thickness, because some terms appear in only one month.)
Two multicolumn tables provide detail about positive and negative terms. For each term, a nested slider (navigationally coordinated with the arc diagram) shows the pattern of monthly occurences as a simple time series. A graph shows selected terms as nodes (blue for positive, red for negative, magenta for mixed). Node size encodes total appearances of each term. Edges connect terms that appear the same month, with thickness representing the number of months in common.
Interaction
The various views allow brushing of positive and negative terms and dynamic filtering on time. Selected terms are highlighted in red in the arc diagram. If the time filter checkbox is selected, the tables and graph filter out terms for months outside the time range visible in the arc diagram. Analysts can brush interesting terms in any of the views, explore patterns of term usage by panning and zooming over time, then drill down to compare temporal patterns for particular terms.
In this screenshot above, the arc diagram reveals a pattern in which appearances of prominent negative review terms increase over the course of a year, suggesting that reviewers may have been reinforcing each others' opinions until the occurrence of a critical event in April 2004.
Downloads
ZIP Visualization (20 KB)
Video Tour (8.0 MB)
References
Chaomei Chen, Fidelia Ibekwe-SanJuan, Eric SanJuan and Chris Weaver. “Visual Analysis of Conflicting Opinions”. Proceedings of the IEEE Symposium on Visual Analytics Science and Technology 2006, Baltimore, MD, October 2006. DOI

PDF Paper (2.3 MB)
Last modified: Mon Feb 18 13:10:51 2008 by Chris Weaver
Copyright © Christopher E. Weaver