Group 1 adopted a very rigid structure based on the use of icons as shown in Figure 1. In this and all the other figures that follow, hierarchical index links are represented as heavy lines between nodes while the lighter lines represent contextual links between nodes. A node is equivalent to an individual file in the document. For these figures, the majority of the links may be considered bi-directional. However, such correspondences had to be explicitly programmed.
In the Group 1 document, each architecture, its properties, capabilities, and environments were identified within the document by group-designed, clickable icons. For example, clicking on the ``capabilities'' icon for the [Laird et al.1987] takes the user to a discussion of the capabilities inherent to that architecture such as learning and problem solving. Each of these architecture sub-topics was made a separate node with links connecting topics within an individual node. Thus, clicking on some property link in the discussion of another architectural property would send the user to a discussion within that same node of the property as it related to the architecture. Each individual topic was identified by a highlighted title which, when clicked, led to a discussion of the topic independent of any architecture. These are shown in Figure 1 as links between the architectural subtopics and the corresponding second-level topic.
As later topics were introduced in the course, this organization seemed particularly unmalleable because such additions required not only the addition of content and new links but the design of new icons and the reformatting of the pages upon which the icons were used. However, the use of icons in identifying subsections within the document made navigation and keeping track of the current location straightforward compared to the other documents.
