Collaborative Research: Decision-making in Collaborative Activities

Harvard University, Cambridge MA

Investigators

Abstract

In the last decade, computer systems have increasingly become participants in complex, distributed communities comprising people and computer systems, rather than devices used by individuals. This major shift engenders a significant challenge for computer science: to determine ways to construct computer systems able to act effectively as team members. Although many current computer systems have sophisticated capabilities as individual actors, most lack capabilities required for working successfully as members of a collaborative group. This proposal addresses problems central to this challenge. It focuses on providing the foundation for the design of decision-making components of software agents that can handle multi-agent, dynamic team contexts. The proposed research comprises three activities: (1) empirical investigations, including examination of the influence on individual and group behavior and outcomes of policies and mechanisms for producing appropriately helpful behavior in collaborative settings and for governing commitment to group activities; (2) development of formal theories that may be used to address abstractly questions of these mechanisms; and (3) construction of more sophisticated collaboration-capable software agents. The research will contribute to the development of collaboration-capable software agents and collaborative human-computer interface systems, and, thus, will significantly increase the effectiveness of heterogeneous teams of people and software-systems agents.