MULTIPARTY PROACTIVE COMMUNICATION: A PERSPECTIVE FOR EVOLVING SHARED MENTAL MODELS
Open Access
- Author:
- Kamali, Kaivan
- Graduate Program:
- Computer Science and Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- April 20, 2007
- Committee Members:
- John Yen, Committee Chair/Co-Chair
C Lee Giles, Committee Member
Jia Li, Committee Member
Kon Well Wang, Committee Member
Raj Acharya, Committee Member - Keywords:
- Distributed AI
Multiagent systems
Agent communication languages (ACL) - Abstract:
- Psychologists who have studied human team performance point out that helping behavior (e.g. proactive sharing of relevant information) in effective teams is achieved via an overlapping “shared mental model” that is developed and maintained by members of the team. In this research, we take the perspective that multiparty proactive communication is critical for establishing and maintaining such a shared mental model among teammates, which is the basis for agents to offer proactive help and to achieve coherent teamwork. Agent communication languages like KQML and FIPA, however, mostly focus on two-party communications and do not provide well defined multiparty performatives. In this research, we provide formal semantics for multiparty proactive performatives within a team setting for agents to share relevant information. The ultimate goal of this research is to develop agent theories and technologies that can assist human decision makers in overhearing multiparty communications, so that they can maintain their evolving shared mental models without being overwhelmed. With this goal in mind, we make the following contributions in addressing the challenges of information sharing among members of a team: First, we propose formal semantics for a set of multiparty proactive performatives in the context of a team. Such performatives can be used in designing agent teams that involve multiparty proactive communication. Second, we formally prove how multiparty proactive communication update the shared mental models among team members, allowing them to proactively offer relevant information to their teammates. Third, we introduce communication policies for multiparty proactive performatives. Finally, we conduct experiments to evaluate the cost and benefits of multiparty communication in a simulated command and control decision making environment for an urban combat application domain.