社会网络影响力传播机制及最优化方法研究

近年来，基于大规模动态社会网络的影响力优化问题成为社会网络领域研究的核心问题。它一方面涵盖积极信息的推广，如影响力最大化；另一方面包括消极信息的抑制，如谣言阻 止。我们统称为影响力传播控制的最优化问题。尽管现有研究给出影响力最大化及谣言阻止的一系列方法，但对二者共享的信息传播机制尚缺少统一深入的优化控制方法。本项目针对社会网络影响力传播控制的最优化问题，首先研究在复杂动态社会网络中决定影响力传播的关键机制，包括节点重要性和节点信息传播能力评估方法，并提出全新的信息传播模型。在此核心传播机制基础上，针对积极信息推广，研究影响力最大化和阈值影响力两类子问题并提出有效近似算法；针对消极信息抑制，研究谣言阻止的K-阻塞者移除问题，K-链接移除问题，K-积极种子选择问题并提出有效近似算法。最后，构建仿真测试平台，对所提出的算法进行系统评估。

In recent years, influence optimization problem has become a hotspot in the study of large-scale dynamic social networks. On one hand, it includes the promotion of positive information, such as Influence Maximization (IM); On the other hand, it also includes the suppression of negative information, such as Rumor Blocking (RB). In this project, we call these problems as the optimization problems of influence propagation control. Although existing research gave a series of methods for IM and RB, in-depth research on the information propagation mechanism and effective optimization control method are still needed. In this project, we first study the key mechanisms of influence propagation in complex social networks including the node importance evaluation and node information propagation capability evaluation. Based on this, we propose new information propagation models. For the promotion of positive information, we propose effective approximate algorithms to solve the influence maximization problem and threshold influence problem, respectively. For the suppression of negative information, we propose effective approximate algorithms to solve the k-Blockers Deletion, k-Edges Deletion and k-Positive Seeds problems. Finally, a simulation test platform is constructed to systematically evaluate the proposed algorithms in this project.