大规模信息传播和情感倾向的实证与分析

项目名称： 大规模信息传播和情感倾向的实证与分析

项目编号： No.91124008

项目类型： 专项基金项目

立项/批准年度： 2011

项目学科： 生物科学

项目作者： 尚明生

作者单位： 电子科技大学

项目金额： 15万元

中文摘要： 非常规突发事件通常会引发谣言，澄清谣言并有效地引导网络舆论具有重要意义。项目在如下三个方面取得重要进展。在实证方面，首先，分析了网络热点事件关注度的时空演变规律，发现关注度的地区分布存在极大的异质性，并与经济和教育水平强关联，舆论引导显著影响事件的关注度，发达地区的舆论引导更有效。其次，分析了短消息系统中人们的行为。在算法方面，鉴于意见领袖在谣言传播或者辟谣过程中的重要作用，提出一种挖掘社会网络中意见领袖的方法，该方法具有准确度高、鲁棒性好、无参数等优点；提出了一种基于邻域信息的半局部中心性识别方法，用于识别无向网络中最具影响力的节点，在速度和效果上取得了很好的平衡；此外，发现在典型社会网络中意见领袖的领导力具有无标度特性，并分析了这种结构的形成机制，发现意见领袖的形成遵从"好者愈富"的机制，即有广泛兴趣和较好判断力的人将成为社会网络中的领导者，对于塑造在线媒体中的意见领袖具有重要参考意义。在模型方面，分析了信息传播和疾病传播的本质区别，提出一个新的消息传播模型，实验表明，当只接收到一次信息就确认的概率较小的时候规则网络的传播比随机网络更快更广，并且小世界能够达到最好的传播效果。

中文关键词： 网络舆情；意见领袖；传播模型

英文摘要： Unconventional emergencies often lead to rumors, and it is of significance to clarify the rumors and effectively guide online public opinion. This project achieved important progress in the following three aspects. First, we empirically analyzed the spatial-temporal evolution of online public opinion and reveal its formation mechanism, found that concerns of regional distribution is great heterogeneous, which strongly correlated with economic and educational levels, and the guide of public opinion has significant impact on the concerns of the event, and guide in developed areas is more effective. Secondly, we analyzed people's behavior in a short messaging system. In the algorithm aspect, first, in view of the important role of opinion leaders, we proposed a method to mine opinion leaders in social networks, which has advantages such as high accuracy, robustness and no parameters; secondly, we proposed a local center identification method based on neighborhood information to identify the most influential to the network node, which achieved a good balance between speed and effect; lastly, we found that in the typical social network of opinion leaders leadership has the scale-free features, and the formation of opinion leaders followed the "richer get richer" mechanism, which means that men of wide interest and good judgment will become a social network. Finally, in the model aspect, we proposed a model to emphasize the essential difference between information spreading and the disease spreading, and experiments show that the spreading effectiveness can be sharply enhanced by introducing a little randomness into the regular structure, namely the small-world networks yield the most effective information spreading.

英文关键词： online public opinion; leadship;spread modeling