动态自适应网络中有界信任舆论演化算法

发布时间：2018-05-05 01:03

本文选题：观点动力学 + 舆论演化　；参考：《西北工业大学学报》2017年03期

【摘要】：目前,对社会舆论演化的研究主要在静态网络中进行。但是,实际的社交关系网络是有向的动态自适应网络,舆论的演化受网络的影响;反之舆论的演化也可能会导致网络的动态变化。对此问题,提出了一种动态自适应网络的有界信任舆论演化算法,其主要功能是:如果2个个体的观点差值大于一定的信任水平,个体之间的有向连接就以一定的概率断开;如果观点差值小于一定的信任水平,新的连接将以一定的概率重新生成。研究的主要内容是:动态自适应网络的平均观点集合数、观点统一概率、最大集合人数比例、平均步数和平均度5种统计指标的宏观变化规律。实验结果表明,最终的观点集合有3种类型:信任水平较小时,形成多个观点集合;中等的信任水平时,以一定的概率达成统一的共识;信任水平较高时,总能达成统一的共识。动态自适应网络中观点演化的趋势和静态网络相似,但在动态自适应网络模型中,重连概率对观点演化有较大的影响。当重连概率为零时,静态网络模型的上述指标优于动态网络模型;当重连概率增加时,动态网络模型的统计指标逐渐优于静态网络模型。重连概率增加了网络的连边,增强了个体之间的交流,促进舆论的快速形成。实验结果与实际网络舆论演化的情况相符,能够在一定程度上反映和解释现实社会中舆论演化的情况。
[Abstract]:At present, the research on the evolution of public opinion is mainly carried out in the static network. However, the actual social network is a dynamic adaptive network, and the evolution of public opinion is influenced by the network. On the other hand, the evolution of public opinion may also lead to the dynamic change of the network. On the evolutionary algorithm, its main function is that if the difference of views between 2 individuals is greater than a certain level of trust, the directed connection between individuals is disconnected with a certain probability; if the difference is less than a certain level of trust, the new connection will be regenerated with a certain probability. The main content of the study is the average view of the dynamic adaptive network. Point set number, viewpoint unified probability, maximum aggregate proportion, average step and average degree of 5 statistical indexes. The experimental results show that there are 3 types of final points of view: the level of trust is small, and a number of viewpoints are formed; when the level of trust is in a certain probability, a unified consensus is reached; the level of trust is the level of trust. The trend of view evolution in dynamic adaptive network is similar to static network, but in dynamic adaptive network model, reconnection probability has a great influence on view evolution. When reconnection probability is zero, the above index of static network model is superior to dynamic network model; when reconnection probability is increased, dynamic network model is better than dynamic network model. The statistical index of the state network model is gradually better than the static network model. Reconnection probability increases the link of the network, enhances the communication between individuals, and promotes the rapid formation of public opinion. The experimental results agree with the actual network opinion evolution, and can reflect the situation of the public opinion evolution in the reconciliation and release of real society to a certain extent.

【作者单位】：西北工业大学计算机学院;西安邮电大学通信与信息工程学院;
【基金】：国家自然科学基金(61271276) 陕西省工业科技攻关项目(2015GY015) 陕西省自然科学基础研究计划项目(2016JM6017) 陕西省教育厅科研计划项目(16JK1702) 西安邮电大学“西邮新星”团队支持计划资助
【分类号】：O157.5

【参考文献】