The Reputation Mechanism In The Public Goods Game And Multi-objective Optimization For Cloud Service Selection

Although the cooperation behavior is a wide spread phenomenon in the human society and nature,the cooperation is usually not an optimal strategy from the perspective of the Darwinian evolutionary theory.Thus,how to explain the persistence and emergence of cooperation has become an active topic within the academia.Among them,the evolutionary game theory has provided a powerful framework to illustrate the evolution of cooperation.With the rapid development of cloud computing technology and the advent of big data era,there are a growing number of available services on the network.Thus,what the users focus on is not the availability of the service,but shifted to the quality of service.At present,users put forward new requirements to the recommendation system.In this thesis,to characterize the influence of reputation mechanism on the spatial public goods game,two new models with reputation are utilized to investigate the evolution of cooperation.In order to meet the requirements of users,we propose two new methods of recommendation to further enhance the diversity of the recommended services.The main contents of the thesis and innovative points are described as follows:1)To put forward an evolutionary game model to investigate the evolution and risk analysis of cooperation under the spatial public goods game(PGG),in which the individual reputation is obviously utilized to cut down the individual risk of being exploited during the evolution of cooperation.In this model,based on the individual utility,the strategy state will be asynchronously updated according to the Fermi-like rule,where the individual utility is characterized as the product of the game payoff and a power function of reputation value.Large quantities of simulations demonstrate that the cooperation behavior can be greatly varied and enhanced when the reputation is incorporated into the utility evaluation.2)To probe into the evolution of cooperation in the spatial public goods game on interdependent lattices by introducing the reputation inferring mechanism into the strategy selection.During the evolution of cooperation,the individual reputation is commonly determined by two corresponding partners on interdependent lattices,where the imitated neighbors are chosen in accordance with the average,maximum and minimum of reputation values between two partners within the neighborhood of a focal player during the strategy update.A large plethora of simulations indicate that three reputation computing rules all lead to the promotion of cooperation when compared to the traditional public goods game model.Among them,the promotion of cooperation under the average and minimum schemes are relatively better than that produced by the maximum rule.The detailed cluster formation and reputation distribution illustrate the slight difference between the outcomes under these three decision making criterions3)To develop two new methods of recommendation to further enhance the diversity of the recommended services and reduce the risk of inappropriate ranking recommendation by adjusting the weight of service origin,under the condition of guaranteeing the accuracy of the suggested service list.Experimental results show that the proposed methods not only can keep the same computational complexity,but also obtain better recommendation results when compared with the original method.