| Mechanism design is the interdisciplinary subject of economic and artificial intelligence.It mainly studies the interaction between agents using game theory.Auction is the typical application of mechanism design,and it also is an effective means of resource allocation.Social network is an indispensable part of modern society,which reflects the interaction and link relationship between social members.It is a problem of essence and frontier to research mechanism design in the social network situation.Besides,such research can make the mechanism design connotation more abundant,make its application scene more diverse.Benefit is always the performance that most mechanism designers are focusing on,yet there are few researches on optimizing the benefit of auction mechanism on the social network.The introduction of social network creates the contradiction between benefit and remuneration of agents.How to balance this contradiction is the difficulty of designing the optimal auction mechanism on social network.This thesis studies the optimal auction mechanism on social network.It is based on the theory of game theory,mechanism design,auction theory and difference constraint system.This dissertation firstly summarizes and expounds the auction mechanisms on social network.It also analyzes the modeling process of auction mechanism on social network.Through the study of agents' behavior,this research reveals the important role of critical nodes for information transmission.In this thesis,the properties of auction mechanism on social network and the reason for the low benefit are summarized.This dissertation also studies the modeling process of auction mechanism on distribution network,which is a special social network.Especilly we study the behavior of intermediaries in the network in the thesis.Intermediaries are unique agents in the distribution netork,who can get commission by providing information to buyers.We also analyzes the existing causes of failure of auction mechanism in distribution network situation.Besides,we promote two new auction mechanims for the single-level and multi-level distribution network respectively.We also prove that these two mechanisms are individual rational,incentive compatibility and weakly budget-balanced.This thesis also models the optimal auction mechanism on social network,and explores the reasons for the difficulty of designing the optimal auction mechanism on social network.The main difficulties are the number of constraints and the solution of functional analysis.By analyzing the characteristics of the incentive compatibility constraints in the optimal auction mechanism on social network,we can simplify the constraints to be a more easy-understanding version,and at the same time the number of incentive compatibility constraints can also be reduced.Based on the constraints of simplification,we revolutionize it into a differential constraint system,which can be solved by using the single source shortest path algorithm on the graph.Finally,it is proved that the payment rules can maximize the benefit of the seller,as long as the allocation rule of the mechanism is satisfied with the monotality. |