Many archives use time-based metadata and keyword tagging to identify documents for later retrieval. However, the conventional browser model offers few possibilities for exploring the patterns that appear when information is classified in this way. Harper’s Weekly Review is one such archive.

Here, a keyword vocabulary evolves over time, reflecting the progress of current events within the context of the magazine’s editorial policy. Until now, the archive has been presented as browser-based hypertext, offering a serendipitous but ultimately somewhat linear experience of its contents. Can a visual tool be developed to understand a data set of such complexity through improved interaction?

An appropriate graphical visualization model allows the discovery of trends and clusters that are otherwise invisible. Meaningful exploration of the data is enabled through manipulation of the information: users make positive choices about how it is displayed, following their own trajectory through interconnected material. The resulting output presents a pictorial landscape that conforms to the meanings inherent in the text.

This talk will examine the development of the Harper’s taxonomy, what lessons were learned from the creation of the visual browser, and how these lessons can be applied to other complex databases.