基于用户相对传播能力的分层网络传播模型 ——以平江火电厂事件为例

An Layering Information Spreading Model Based on Transmission Capability in Social Network ——A Case Study of Pingjiang thermal power event

摘要:

根据社交网络信息传播多维度、多元化和基于人际关系的特点，构建一种基于用户相对传播能力的分层网络传播模型，对社交网络传播模型研究中缺少实例验证和用户层次分类的问题进行了补充分析和证明。通过调查收集“平江火电厂重启事件”中公众对火电项目的态度变化数据，对政府相关宣传内容的传播过程、方式和效果进行归纳总结；采取K-Means方式对抽样的数据节点进行分类，构建以节点传播能力为划分标准的分层网络传播模型，分析模型中信息的传播路径、传播特征及权重关系。研究表明：（1）基于用户相对传播能力的分层网络传播模型更能真实地体现社交网络中信息传递的特点；（2）意见领袖层节点在信息传播过程中具有导向性作用，其网络影响程度约为普通群众的4倍（意见领袖节点的平均权重为2.6，普通群众节点的平均权重为0.6）；（3）用户网络地位影响信息传递的效果，政府在提高公众支持度的问题上应该充分发挥意见领袖的作用，引导社会舆论。社会信息传播网络仍然是复杂网络研究中的难点，信息传递的方向、节点出度和入度都有待在未来的研究中进一步完善。

Abstract:

In this paper, we first introduce information spreading in social networks characterized by multi-dimensional and diversity based on human relations. Then, we propose a layering information spreading model based on relative weight, and analyze questions about short of instance and user segmentation. Taking Pingjiang thermal power event as the example, we transfer data from government propaganda with node classification by K-Means. By analyzing influence on different paths and relative weight, we find that the information is sent by the government, then it is spread more and more broad through the opinion leader layer, and everyone finally receives the information faster and more comprehensively. The results show that (1)The model of layering information spreading can reflect the real social network characteristics and the information spreading is influenced by the status of spreading nodes. (2)The opinion leadership node is the guided section in network, and its network effect is about four times as large as ordinary people (The average weight of opinion leader node is 2.6, which is much larger than ordinary ones owning 0.6 ). (3)The government may give full play to the role of opinion leaders in improving the public support and guiding public opinion. Social information transmission network is the research issues in complex networks. Therefore, we will improve the transmission direction and the degree of nodes in the future study.