Research And Implementation Of Task Oriented On Team Formation Mechanism

With the development of artificial intelligence,the application field of intelligent agent is expanding,and the work of agent is becoming more complicated.Because of the limited ability of a single agent to work,the multi-agent team has become a fast and efficient form for solving complex problems.Multi-agent teams accomplish complex tasks through cooperate the complementary skills of agents.Team members share skills of tasks and resources at work,in the form of team organization.Many scholars at home and abroad have improved the selection methods for team members,and attempted to conduct breakthrough research to find more practical selection methods.Based on this the main work and specific contributions of this article are as follows:A robust mechanism for agent team formation has been proposed,on the issue of team formation with single task.The mechanism designs a SC algorithm to select team coordinator,the team coordinator sends offer to agents,and negotiates with agents about whether the agent contributes abilities to the team,to realize the intent that can make joint decisions.At last the mechanism calculates the robustness of teams to select the optimize team,which has the highest team robustness and the least cost.An example is given to illustrate the robustness of the mechanism,and the experiment shows that the team's cost will not increase when increasing the team's robustness.A TFCF mechanism and a TFLD mechanism have been proposed,on the issue of team formation with multi-task.The TFCF mechanism selects the agents who has great contribution to the task and has the skill that the task requirements,according to the order of the working strength of task.The TFLD use the shortest path generation algorithm in undirected graph to select agents,leader makes a comprehensive plan to the task,according to the skill requirements and time requirements of the task.Compared with the existing multi-task team formation mechanism,the contribution priority team formation mechanism improves the utilization of agent's effective skills,and the leader decision team formation mechanism ensures the efficient completion of the task by the comprehensive plan.A subsection utility distribution mechanism(SDM),a truthful payment mechanism(PMC)to the buyer,and a self-protection pricing function to the seller have been proposed,on the issue of internal utility sharing for teams.The SDM and the PMC are based on contribution formed by the max-utility of team and the max-deal of team.In SDM,buyer's utility is proportional to the contribution of the team.In PMC,the payment of each buyer is its contribution to the team,and the self-protection pricing function protects the utility of seller.The theoretical analysis reveals individual rationality,budget balance,truthfulness and fairness of the two mechanisms.Experimental results show that the self-protection pricing function satisfies the nature of the increase of the seller's income,and can guarantee that the seller's income does not decease with the increase of the goods selling.