Measuring Text-Graphic Activity

Fred Lakin

Rehabilitation R&D Center, Palo Alto Veterans Hospital
3801 Miranda Ave, Palo Alto, California 94304

Interface designers and psychologists need to be able to formulate hypotheses about user activity in graphics interfaces and then experimentally test them. The results of such experiments are both interesting in their own right and can be used to change the design of interfaces so as to better support the graphics activity taking place. In order to do this, what is needed is a high level representation for the activity of graphics interface use. In terms of such a representation: 1] hypotheses can be described; 2] interface activity can be measured to test the hypotheses; and perhaps even 3] changes to the interface design can be specified. This paper presents a structural model of writing and drawing which provides a metric for measuring such activity and is designed to serve the three purposes above.

In the SAM model, the product of writing and drawing is simplified to text-graphic objects, and the activity of writing and drawing becomes text-graphic manipulation. The text-graphic objects have structure. This structure arises directly from an attempt to account for the manual manipulations observed in non-computer image production such as occurs on blackboards. According to the model, the needs of manual manipulation determine the text-graphic pattern as the simplest organizing structure for images. SAM stands for Structure-Arises-out-of-Manipulation. Included in the SAM model is a notation for the structure of text-graphic objects. This notation allows high level description of blackboard type image activity. And because the notation has been implemented as a programming language, the text-graphic objects can be computed with, grounding such measurement-useful capabilities like processing, parsing and evalution.

A graphic structure editor based on the SAM model has been defined. The editor was implemented in PAM, a language which generalizes LISP to handle text-graphic objects (PAM stands for PAttern Manipulation).

The model-based editor has been used to provide measurement of and interactive assistance for text-graphic manipulation. The simplest measurement is simply a chronological record of each successive manipulation and image state. The lowest level assistance is structure based agility aids. Next, direct user manipulation of structure is facilitated. And at higher levels, the editor is a tool for exploring the rules used by humans to collect elementary visual objects into conceptual groups. Examples of such groupings discovered by analysis of user editor activity are presented in the paper.

Implementations have been done in MACLISP, Smalltalk and Allegro Common LISP.

KEYWORDS: graphics interface, measuring text-graphic activity.

Transcription of first measure of paper performance by Sibbet


Structure of and from Manipulation in the image


Attention shifting by levels during image construction



Blackboard Activty is Prototypical of Writing and Drawing
An Example of Blackboard Activity
Chunking the Objects in Blackboard Activity
The Purpose of Measuring Writing and Drawing Activity

The SAM Model: Structure Arises out of Manipulation
Consequences of the Sam Model
Grouping Structures for the Blackboard Image

Embodiment of the Model: handPAM, a Graphics Editor
Image Example: the Project Diagram
Generalizing LISP to Visual Objects
Measurement in the handPAM Environment
Measuring Structures in the 'Project Diagram' Activity
Dynamics of Image and Structure



Published in the proceedings of GRAPHICS INTERFACE '83, Edmonton, Alberta, May 1983.

Return to papers list

(C) Copyright 1994 PGC