社会网络中竞争与合作影响最大化研究

发布时间：2018-05-20 18:42

本文选题：社会网络 + 竞争影响最大化　；参考：《云南大学》2016年博士论文

【摘要】：随着在线社交网络的流行,社会网络影响传播引起很多研究者的关注,社会网络影响传播最大化是社会网络影响传播的关键问题之一。社会网络影响传播最大化是指给定一个社会网络、传播模型和种子节点个数,找到指定个数的种子节点集合,使得影响通过这个种子节点集合传播最大化。社会网络影响传播最大化在病毒式营销和信息的传播等方面具有重要应用。在病毒式营销和信息的传播中,不仅存在单一的影响传播,还存在竞争和合作影响传播。本文针对社会网络中病毒式营销和信息传播的应用背景,研究了社会网络中竞争与合作影响最大化,并用实验验证了本文方法的可行性和有效性。本文的主要工作和创新之处总结如下：●研究了社会网络影响传播中种子选择的并行方法。本文基于候选壳生成、热扩散模型及候选壳影响最大化的基本思想,提出了基于候选壳的并行选取种子节点集合最大化社会网络影响传播方法。该研究以提高选取种子的速度为出发点,基于候选壳,可并行地在候选壳中选取种子节点集合；基于热扩散模型模拟病毒式营销中影响的传播,引入时间参数,从而更好地模拟病毒式营销中影响的传播。●研究了面向病毒式营销的社会网络竞争影响最大化。本文基于扩展线性阈值模型、子模性分析框架及贪心法的基本思想,提出了病毒式营销中竞争影响传播最大化的方法。该研究以社会网络的实际应用为出发点,基于扩展线性阈值模型模拟竞争影响传播,从而更符合实际地模拟病毒式营销中的竞争影响传播；基于子模性的分析框架,为贪心法近似地选取种子节点集合提供了理论保证。●研究了面向信息传播的社会网络竞争影响最大化。本文基于可能图、竞争影响传播模型、子模性分析框架和贪心法改进算法的基本思想,提出了信息传播中竞争影响最大化的方法。该研究以可能图为出发点,大大消除了计算的难度,提高了选取种子节点集合的速度；采用竞争影响传播模型来模拟可能图中的竞争影响传播过程：基于子模性的分析框架,从而可以采用花费有效的懒惰向前算法来近似地选取种子节点集合,花费有效的懒惰向前算法是一种贪心法的加速算法。●研究了面向病毒式营销的合作影响传播最大化。以模拟病毒式营销中商品的合作促销为目的,本文基于相似度模型、合作影响传播图、带接受概率的独立级联模型和改进的贪心算法的基本思想,提出了病毒式营销中合作影响传播最大化的方法。该研究以面向病毒式营销的合作影响传播最大化为出发点。基于单独一种商品的影响传播图及关联规则生成合作影响传播图；采用带接受概率的独立级联模型来更符合实际地模拟病毒式营销中合作影响传播过程；提出改进的贪心法来有效地选取种子节点集合,避免了大量的蒙特卡洛模拟计算。
[Abstract]:With the popularity of online social networks, social network impact communication has attracted many researchers' attention. The maximization of social network impact communication is one of the key issues of social network impact communication. The maximization of social network influence propagation means that given a social network, the propagation model and the number of seed nodes, the set of seed nodes is found to maximize the propagation of influence through the set of seed nodes. The social network influence communication maximization has the important application in the virus marketing and the information dissemination and so on. In viral marketing and information dissemination, there is not only a single impact of communication, but also competition and cooperation. Aiming at the application background of viral marketing and information dissemination in social network, this paper studies the maximization of the influence of competition and cooperation in social network, and verifies the feasibility and effectiveness of this method by experiments. The main work and innovations of this paper are summarized as follows: the parallel methods of seed selection in social network influence propagation are studied. Based on candidate shell generation, thermal diffusion model and the basic idea of maximizing the impact of candidate shells, this paper proposes a parallel selection of seed node sets based on candidate shells to maximize the social network impact propagation method. Based on the candidate shell, the seed node set can be selected in parallel, based on the thermal diffusion model to simulate the spread of the influence in viral marketing, and the time parameter is introduced. In order to better simulate the spread of influence in viral marketing, this paper studies the maximization of the impact of social network competition for viral marketing. Based on the extended linear threshold model, the submodule analysis framework and the greedy method, this paper proposes a method to maximize the spread of competitive influence in viral marketing. This study takes the practical application of social network as the starting point and simulates the spread of competition influence based on extended linear threshold model, which is more in line with the actual simulation of competition impact propagation in viral marketing, and based on the analysis framework of submodule. It provides a theoretical guarantee for the greedy method to select the seed node set approximately, and studies the maximization of the competitive impact of the social network oriented to information dissemination. Based on the basic ideas of possibility graph, competition influence propagation model, submodule analysis framework and greedy method, this paper proposes a method to maximize the influence of competition in information dissemination. In this study, the difficulty of calculation is greatly eliminated, and the speed of selecting seed nodes is improved. The competitive influence propagation model is used to simulate the process of competition influence propagation in the possible graph: an analysis framework based on submodule. Therefore, the cost effective lazy forward algorithm can be used to approximate select the seed node set. The cost effective lazy forward algorithm is an accelerated algorithm of greedy method. For the purpose of simulating cooperative promotion of merchandise in viral marketing, this paper bases on similarity model, cooperative influence propagation graph, independent cascade model with acceptance probability and improved greedy algorithm. This paper puts forward the method of maximizing the influence of cooperation in viral marketing. This study is based on the cooperative impact of viral marketing on communication maximization. Based on the influence propagation diagram and association rules of a single commodity, the cooperative impact propagation map is generated, and the independent cascade model with acceptance probability is used to simulate the process of cooperative influence propagation in viral marketing more accurately. An improved greedy method is proposed to select seed nodes effectively, avoiding a large number of Monte Carlo simulations.
【学位授予单位】：云南大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：G206;TP393.09

【参考文献】