This proposal addresses problems central to the design of computer
agents that can negotiate proficiently with people in strategic
settings. Due to advancing technology heterogeneous groups that
include both people and computer agents making decisions together are
becoming prevalent. Despite the numerous benefits of computer-aided
negotiation, the design of agents that negotiate with people has
received less attention than the design of agents for multi-agent
settings that comprise solely computer agents. This proposal focuses
on computational strategies for information exchange that can
facilitate human-computer negotiation by allowing people to change
their positions. The research program will
consist of settings that vary in complexity across multiple
dimensions, including the expressiveness of the argumentation
protocol, the availability of information about participants'
preferences and goals, the extent to which participants can renege on
past commitments during the negotiation, and the number of rounds of
negotiation. For each of these settings, I will develop computational
strategies by using a variety of theories and algorithms. I will
compare the performances of agents using these different strategies
and their effect on people's behavior. The empirical studies will
be directed at determining the factors that influence people's
negotiation behavior and at testing the performance of different agent
strategies.