Simply put, a graphic dialog occurs when two entities are trading text-graphics back and forth. entityOne creates some text-and-graphics, and, by putting them on the screen, submits them to entityTwo, who looks at the text-and-graphics, and then returns some text-and-graphics which entityOne looks at, and so on and so forth. In this case, one entity is the human and the other is a visual agent.
When a visual agent is assisting group graphics, the human generates text-and-graphics representing the group's cognition, and the agent intervenes on the display from time to time. For ease of reference, the visual agent for group graphics has been named "Dave." When Dave goes into action, he appears on the screen as an eye (for seeing) and an arm with hand (for manipulating). Dave's intervention is triggered when the human creates certain visual patterns in the course of group graphics.
Patterns Dave currently recognizes include arrangements involving lists of textual items with hand-drawn bullets; pieces of text connected to any text-graphic object by a hand-drawn straight arrow; two text-graphic objects connected by straight lines (figure 2); and hand-drawn hollow arrows, with text inside and at the tail end, pointing to any text-graphic object.
These patterns are used in a visual language developed for group facilitation by a talented graphical performer [Sibbet80]. When the human creates a pattern which is recognized, Dave then generates a text-graphic response and this object is displayed on the screen. The response is appropriate both to the general type of visual pattern which triggered it, and to specific elements in each individual triggering pattern (if part of the triggering pattern is an element that Dave cannot deal with, such as an arbitrary cartoon, then that element is simply copied directly into the response using a template based generation technique).