Computational Models for Mixed Initiative Interactions
Papers from the AAAI Spring Symposium
Susan Haller & Susan McRoy, Cochairs
In a mixed initiative interaction, direction and control of the interaction shifts among the participants. Since the information and abilities needed to solve a problem are distributed among the participating agents, a system that is collaborating with other users to solve a problem must have the flexibility to take or relinquish initiative. However, the possibility of shifting initiative requires that computer systems include mechanisms for recognizing when to lead or otherwise take control of an interaction and when to let others take the initiative. In addition, an interactive system must be able to take initiative into account in interpreting utterances and in selecting appropriate responses. This symposium brought together researchers who were interested in developing theoretical and applied models for mixed-initiative interac-tion. As a group, participants discussed core issues such as the factors that characterize initiative, and the role that initiative should play in the design of computer systems. Specifically, they addressed the following issues: (1) What knowledge strategies or knowledge representation schemes (e.g. plans, logics, etc) are appropriate for modelling initiative. (2) What algorithms or computer systems can be developed for controlling mixed-initiative in-teraction. This area of research is just reaching the point where the central issues can be discussed within a focused community. Identifying and establishing this community is an appropriate and key function of this AAAI symposium.