社会网络结构是影响社会行为的重要因素。舆论形成作为经典的社会集群行为，已有一些用复杂网络的方法以及应用Ising 模型来研究其集群动力学的成果。然而，将Ising 模型应用于符号网络上对舆论形成进行的研究却很少。本文以舆论形成为研究对象，主要关注带有正负的符号网络，并对Ising 模型加以修正使其适用于异质性相互作用的网络。分别以人工符号网络和实际符号网络为例，探究在不同的初始化条件：只有正边和有不同负边比例的网络结构，对舆论传播过程的影响，以及修正后Ising 模型在符号网络上呈现的新特性。另外，在实际网络中，本文还研究了网络达到各种状态的稳态概率与系统能量之间的关系。

The social network structure is an important factor influencing social behavior. As a classic social cluster behavior, the formation of public opinion has some methods using complex networks and the application of the Ising model to study the results of its cluster dynamics. However, it is very rare to apply the Ising model to the formation of public opinion on the signed network. This paper takes the formation of public opinion as the research object, mainly focuses on the positive and negative sign networks, and modifies the Ising model to make it suitable for the networks of heterogeneous interaction. Respectively, take the artificial signed network and the actual signed network as examples to explore the different initialization conditions: only positive side and different proportion ofnegative side of the network structure, the influence on the propagation process of the public opinion, and the new features that modified Ising model presents on the signed network. In addition, in the actual network, this paper also studies the relationship between the steady-state probability and the system energy of various states.