基于社交情境分析的虚假信息用户传播意图检测及使用控制

社交网络虚假信息传播研究可以追溯到早期复杂网络和小世界网络中的谣言传播动力学模型，直到近三年来关于社交用户和社交机器人两类主体的混合型、交互式传播模式和行为研究。然而，基于这两类主体的传播行为，研究其潜在的、隐藏的传播意图检测及使用控制等，还鲜有文献报道。本项目创造性地融合情境分析（计算）和使用控制（安全）理论，探索行为—意图分析的计算理论和安全控制机制，旨在解决日益突出的虚假信息随意传播问题。通过构建社交情境分析框架，提出基于传播行为特征分析的半监督聚类算法，用于区分两类主体；进而，着重研究社交用户的传播意愿推理和转移，实现虚假信息传播意图检测和传播趋势目标识别；提出具有时空情境特征的细粒度使用控制模型和机制，实施事前和事中及时有效的传播控制；最后在一个现实社交网络平台开展实验和验证。研究将对建立信息传播的社交情境分析可计算理论，实现社交网络空间安全，具有重要基础理论意义和广泛应用前景。

Research on fake information propagations in social networks dated back to exploring dynamics models on the rumours spreading in complex network and small-world network previously. Even within recent three years, some hybrid and interactive propagation pattern and behavior studies on two kinds of entities of social human users and socialbots have been done. But, there is rare literature report on research works on their potential and hidden propagation intentions detection and usage control based on the above propagation subjects’ propagation behaviors. This project has a creative convergence between situational analytics (computing) and usage control (security) theories to explore the behavior-intention analysis computational theory and security control mechanism, with the aim at resolving the burning issue of the fake information propagation at user’s will. By establishing a novel social situational analytics framework, a semi-supervised clustering algorithm would be proposed based on the propagation behavior features analysis, in order to classifying two kinds of entities. Further, the propagation desire inference and transition are explored, implementing the social users’ propagation intention detection and distribution trends and goals identification. And then, a fine-grained temporary-spatial usage control model and mechanism are proposed for the beforehand and run-time propagation control efficiently and effectively. Finally, the experiments and verifications are made by using a real-world social network platform. The study has fundamental theoretical significance for establishing the SocialSitu analytics computational theory for information propagations, and wide applicable visions for establishing social networks security.