社交媒体上用户行为驱动的信息传播过程的建模、实证及应用

Online social media has become the main tools for people to spread information. Combining the complex real behaviors of users to accurately understand the rules of information dissemination, it is not only conducive to the spread of new ideas and new products, but also can improve the supervision of public opinion on social media. In recent years, information spreading on online social networks has attracted much attention from academia and gradually formed a new discipline--computational communication. However, since it is relatively difficult to obtain the entire network structure and communication tracks of the social media, the data-driven model established in this field lacks a detailed empirical comparison with the actual disseminating information, so their effectiveness and accuracy are questionable. In view of this, the project intends to study the following issues in depth step by step: 1) based on a large number of empirical data, the user behavior is integrated into the social media networks, and spreading models embedding of behavior factors are set up; 2) compare the actual information with the established spreading model, extract the key behavioral factors, such as user activity and diffusivity differences, and establish appropriate models; 3) put forward targeted vital user identification algorithm based on user behavior characteristics and specific rules of information disseminating. The research results of this project will improve the empirical study of the spreading model, so as to deepen the understanding of the mechanism of dissemination and provide a theoretical basis for the important applications of social media in rumor control and social sensing.

在线社交媒体已经成为人们传播信息的主要手段，结合用户复杂的真实行为来准确认识信息的传播规律，不仅有助于传播新思想与新产品，还可以提高社交媒体上舆论监管效率。近些年，社交网络上的信息传播备受学术界关注，并逐渐形成一门新的学科--计算传播学。然而，由于完整获取整个社交网络结构与传播轨迹相对困难，一直以来，该领域建立的数据驱动模型缺少与实际信息传播的实证精细对比，因而有效性和精确性有待商榷。鉴于此，本项目拟逐步深入地研究以下问题：1）基于大量实证数据，将用户行为融合到社交媒体网络，并建立行为因素内嵌的传播模型；2）对比实际信息与建立的传播模型，提取关键行为因素，如用户活跃度、传播性差异，构建合适的模型；3）基于用户行为特征及信息传播的特有规律，提出有针对性的关键用户识别算法。该项目的研究成果将完善传播模型的实证研究，从而加深对传播的机理认识，为社交媒体在谣言控制、社会感知等重要应用提供理论基础。

在线社交媒体已经成为人们传播信息的主要手段，结合用户复杂的真实行为来准确认识信息的传播规律，不仅有助于传播新思想与新产品，还可以提高社交媒体上舆论监管效率。然而，由于获取整个社交网络结构与传播轨迹相对困难，一直以来，该领域建立的数据驱动模型缺少与实际信息传播的实证精细对比，因而有效性和精确性有待商榷。项目基于大规模社交网络的实际数据，建立了社交媒体信息传播的数据驱动模型，和社交网络次近邻作用机制的传播模型。项目提取了用户行为与网络结构共演化的关键因素，据此提出了新的传播模型，能很好地解释社交媒体的强传播能力，更准确预测信息的传播，验证了过往关于信息传播的渗流相变理论。并发现社交媒体关键节点影响力的极度不均衡现象，为社交媒体在关键用户识别、谣言控制、社会感知等重要应用提供理论基础。进一步，项目以科学家社交网络为例，研究了在新领域兴起时科学家的研究领域转变行为，并提出了具有次近邻作用机制的传播模型。