复杂网络下博弈机制与信任预测的研究

发布时间：2018-07-08 21:05

本文选题：复杂网络 + 公共物品博弈　；参考：《天津理工大学》2017年硕士论文

【摘要】：复杂网络的出现,为复杂性研究提供了一种新视角与新方法,它以一种抽象的方式描述真实的网络。与此同时,演化博弈论的引入为人们解决社会困境问题提供了强有力的理论工具,两者相结合进一步拓宽了人们对演化合作行为的理解。论文主要依据复杂网络理论,使用建模仿真方法,研究囚徒困境博弈、雪堆博弈、公共物品博弈模型中群体的合作演化行为;利用复杂网络建模思维,对在线社交网络中的信任预测进行理论建模与实验。论文的主要研究成果以及创新点包括以下几个方面:(1)提出一种新的自相互作用机制,将其应用到结构化网络中,研究自相互作用在囚徒困境博弈与雪堆博弈中对演化合作行为的影响。基于囚徒困境博弈与雪堆博弈,考虑多种不同的自相互作用强度对群体合作行为的影响。另外,在自相互作用博弈机制的基础上,研究个体差异性对群体合作行为的影响。数值仿真结果表明,自相互作用博弈机制的引入能够大大促进结构化群体中的合作水平。研究结果有助于帮助人们有效地配置各种资源,避免陷入无效率境地,并且有利于解决诚信建设问题。(2)探索个体差异性与投资异质性对公共物品博弈的影响。模型假设两种类型的个体,分别为A类型与B类型,通过控制A类型与B类型个体在群体中的比例达到个体差异目的,并用参数?调节投资异质程度。仿真模拟结果表明,A类型个体与投资异质性对于群体合作具有积极作用,可以大幅提升群体的合作水平。研究结果为人们解决社区公共服务的提供、全球温室气体的减排等提供了良好的理论框架与方案。(3)研究利用复杂网络建模思维,对社交网络中信任关系进行相关预测。将多种社会因素融合进行建模,采用非监督学习方式预测用户之间的信任关系。利用公开数据集Epinions进行的初步实验结果表明,将社会因素中的同质性与时间效应引进模型中可以适当地提高预测准确率,提高了解信息的受众范围,使得在线网络中的信息传递更加有效。
[Abstract]:The emergence of complex networks provides a new perspective and new method for complexity research. It describes real networks in an abstract way. At the same time, the introduction of evolutionary game theory provides a powerful theoretical tool for people to solve social difficulties. The combination of the two further widens people's understanding of evolutionary cooperative behavior. Based on the complex network theory, the thesis uses the modeling and simulation method to study the cooperative evolution behavior of the group in the prisoner's dilemma game, the snow pile game, the public goods game model, and the model and experiment of the trust prediction in the online social network. The following aspects include: (1) a new self interaction mechanism is proposed, which is applied to a structured network to study the influence of self interaction on the evolutionary cooperative behavior in the prisoner's dilemma game and the snow pile game. Based on the prisoner's dilemma game and snow pile game, a variety of different self interaction intensities have been considered for the group cooperative behavior. In addition, the influence of individual diversity on group cooperative behavior is studied on the basis of the self interaction game mechanism. The numerical simulation results show that the introduction of self interacting game mechanism can greatly promote the level of cooperation in structured groups. The results are helpful to help people to effectively configure various resources and avoid being ineffective. (2) explore the effect of individual difference and investment heterogeneity on the game of public goods. The model assumes that two types of individuals are A type and B type, respectively, by controlling the proportion of A type and B type individual in the group, and using parameters to adjust the investment heterogeneity process. The simulation results show that A type individual and investment heterogeneity have positive effect on group cooperation, and can greatly improve the level of group cooperation. The results provide a good theoretical frame and scheme for people to solve community public service and global greenhouse gas emission reduction. (3) research and use of complex network modeling thinking To predict the relationship between the social networks and the relationship between the social networks. Modeling a variety of social factors and using the unsupervised learning method to predict the trust relationship between users. The preliminary experimental results using the open data set Epinions show that the introduction of the homogeneity and time effect in the social factors can be improved properly. The accuracy of prediction can be improved, and the audience of information will be improved, so that the information transmission in online network will be more effective.
【学位授予单位】：天津理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O157.5

【相似文献】