Exploring Browser Design Trade-Offs Using a Dynamical Model of Optimal Information Foraging

Designers and researchers of human-computer interaction need tools that permit the rapid exploration and management of hypotheses about complex interactions of designs, task conditions, and user strategies. Dynamic programming is introduced as a such a tool for the analysis of information foraging technologies. The technique is illustrated in the context of the Scatter/Gather text clustering browser. Hypothetical improvements in browser speed and text clustering are examined in the context of variations in task deadlines and the quality of the document repository. A complex and non-intuitive set of tradeoffs emerge from even this simple space of factors, illustrating the general utility of the approach. Keywords Dynamic programming, information foraging, Scatter/Gather, user models.

• Type: Conference
• Year: 1998
• Where: CHI
• Authors: Peter Pirolli
• Keywords: CHI 1998, Dynamic Programming, Human Computer Interaction, Information Foraging, Text Clustering