Dynamic management of multichannel interfaces for human interaction with computer-based intelligent assistants.

• Main Author: Strickland, Ted John, Jr.
• Other Authors: Ferrell, W. R.
• Language: en
• Published: The University of Arizona. 1989
• Subjects: Human-machine systems.
• Online Access: http://hdl.handle.net/10150/184793

For complex man-machine tasks where multi-media interaction with computer-based assistants is appropriate, a portion of the assistant's intelligence must be devoted to managing its communication processes with the user. Since people often serve the role of assistants, the conventions of human communication provide a basis for designing the communication processes of the computer-based assistant. Human decision making for communication requires knowledge of the user's style, the task demands, and communication practices, and knowledge of the current situation. Decisions necessary for effective communication, when, how, and what to communicate, can be expressed using these knowledge sources. A system based on human communication rules was developed to manage the communication decisions of an intelligent assistant. The Dynamic Communication Management (DCM) system consists of four components, three models and a manager. The model of the user describes the user's communication preferences for different task situations. The model of the task is used to establish the user's current activity and to describe how communication should be conducted for this activity. The communication model provides the rules needed to make decisions: when to communicate the message, how to present the message to the user, and what information should be communicated. The Communication Manager controls and coordinates these models to conduct all communication with the user. Performance with DCM as the interface to a simulated Flexible Manufacturing System (FMS) control task was established to learn about the potential benefits of the concept. An initial comparison showed no improvement over a keyboard and monitor interface, but provided performance data which exposed the differences in information needed for decision making using auditory and visual communication. This knowledge and related performance data were used to redesign features of the DCM. The redesigned DCM significantly improved all aspects of system performance compared to the keyboard and monitor interface. The FMS performance measures and performance on a secondary task improved, user communication behavior was changed favorably, and users preferred the advanced features of DCM. These types of benefits can potentially accrue for a variety of tasks where multi-media communication with computer-based intelligent assistants is managed with DCM.