9360764 Zachary Computerization of work, particularly information-based work which has been traditionally performed by manual methods, has created many problems for new computer users. As computer systems grow more complex, their human users have increasingly difficulty in successfully or efficiently applying the underlying system functionality. Research has begun to focus on using the human- computer interface to solve such problems by creating components of the user interface, called `agents' or `interface agents', that cooperatively support the human in a variety of ways. To date, however, these interface agents have been the custom-built products of individual research or system-development efforts. For the concept and benefit of interface agents to reach its broader potential, more powerful tools are needed for their development and maintenance. This research proposes to assess the feasibility of creating such a Generator of INterface Agents (GINA). The proposed tool is developed around an interface agent-building methodology called COGNET, that was created, successfully applied and behaviorally validated by the Principal Investigator in prior research. The proposed research develops and analyzes a detailed GINA architecture, then implements and tests a skeletal version of the underlying agent-execution engine.
