Investigating and improving network visualizations' effectiveness at supporting human sensemaking tasks

Network visualizations are being used to display relationships in a variety of domains. This is due to their almost universally applicable intended purpose of helping viewers to understand components, attributes, and interconnections of virtually any type of system or environment. Despite their growing use however, there is little empirical research on what types of tasks these visualizations can support and how to design them to support most effectively these tasks. This paper presents findings from an empirical study focused on determining if and how network visualizations can be designed to support information foraging tasks. Results showed foraging performance increased when supported by a network visualization integrating visual cues representing attributes of represented entities and relationships. The largest improvements were in situations where searches involved multiple relevant information categories. There was almost no improvement in performance and in some cases a decrease, compared to a text-only condition, when a basic network display without attribute specific information was used.