Abstract: A large number of studies have found that the social network structure has a significant influence on the dynamic characteristics which occur on itself. The DNCSN（Social Network Generation Algorithm with the Constraint of Dunbar’s Number） proposed in this paper has a small average distance and a large clustering coefficient, and can generate results close to the real network on the nonlinear relationship between the degree and strength of nodes. It can be more effective to study on the characteristics of opinion evolution based on DNCSN. This paper compares the changes in the number of echo chambers formed in different network conditions by adjusting the opinion “firmness” and opinion confidence threshold parameters of each individual in the social network. The results show that when the initial opinion of the network is evenly distributed and the size of network is less than Dunbar’s number, the structure of DNCSN can reduce the number of echo chambers, compared with fully connected network of the same size, within the range of the confidence threshold is larger than 0.15. When the scale of network increases, the DNCSN can form more echo chambers than fully connected network. When the range of the confidence threshold is less than 0.1, the scale of fully connected network will affect the quantity of echo chambers and the smaller the network size is, the smaller the number of echo chambers will be.

Key words: opinion dynamics, social network, network generation algorithm, echo chambers, Dunbar’s number

摘要： 大量研究发现，社交网络结构对观点的传播动力学特性有显著影响。提出的基于邓巴数限制的网络生成算法——DNCSN（Social Network Generation Algorithm with the Constraint of Dunbar’s Number），具有较小的平均距离和较大的群聚系数，可以生成与真实网络节点度和节点强度的非线性关系较为接近的结果。基于DNCSN生成算法，可以更为有效地研究社交网络上的观点演化特性。通过调整社交网络中个体的观点“坚定度”和观点取信阈值参数，对比研究了回音室的形成数量变化。研究表明当网络的观点初值均匀分布，网络规模小于邓巴数字时，相比于同等规模的全联通网络，DNCSN网络结构能在节点取信阈值大于0.15时减少网络的回音室的形成数量；当网络规模增大时，DNCSN网络的回音室形成数量多于全联通网络；当取信阈值小于0.1时，全联通网络的回音室形成数量受网络规模影响，且网络规模越小回音室数量越少。

关键词: 观点动力学, 社交网络, 网络生成算法, 回音室, 邓巴数