¹Computer Science Department & Human–Computer Interaction Lab, University of Maryland, College Park, MD, U.S.A.

Correspondence: Ben Shneiderman, Computer Science Department & Human–Computer Interaction Lab, University of Maryland, College Park, MD 20742, U.S.A.. Tel: +1 301 405 2680; Fax: +1 301 405 6707; E-mail: ben@cs.umd.edu

Received 27 July 2007; Revised 2 October 2007; Accepted 3 October 2007; Published online 22 November 2007.

Abstract

A semantic substrate is a spatial template for a network, where nodes are grouped into regions and laid out within each region according to one or more node attributes. This paper shows how users can be given control in designing their own substrates and how this ability leads to a different approach to network data exploration. Users can create a semantic substrate, enter their data, get feedback from domain experts, edit the semantic substrate, and iteratively continue this procedure until the domain experts are satisfied with the insights they have gained. We illustrate this process in two case studies with domain experts working with legal precedents and food webs. Guidelines for designing substrates are provided, including how to locate, size, and align regions in a substrate, which attributes to choose for grouping nodes into regions, how to select placement methods and which attributes to set as parameters of the selected placement method. Throughout the paper, examples are illustrated with NVSS 2.0, the network visualization tool developed to explore the semantic substrate idea.