ATD: Algorithms for Threat Detection in Knowledge Graphs

该项目开发了涉及知识图谱的新的数学算法和模型。知识图以带标签的节点和边的形式表示关于一个主题的已知内容。知识图不仅仅是简单地标记数据，它根据高级含义组织数据，并为图中的每个节点分配全局唯一标识，以匹配现实世界的实体。知识图谱的很多工作都是处理数据库和查询的。相比之下，在威胁检测的背景下，本项目侧重于识别图中潜在信息的算法以及与图上数据相关的预测模型。该项目将涉及子图同构检测、时间序列分析、基于代理和多尺度建模以及模式识别的数学方法的组合。本项目将通过参与研究培养1名博士后学者、1名博士生和6名本科生。这个项目汇集了几个不同的重点问题与大型，多模态，复杂的数据集。数据被组织成一个知识图，其中添加和吸收额外的信息，因为它变得可用。本项目考虑了三种不同应用类型的知识图谱：(1)由复杂的多部分叙事构建的知识图谱；(2)基于异构网络内容构建的知识图谱；(3)与大规模人类互动动态（如全球流行病）相关的知识图谱。对于(1)，将设计算法来识别重要的因果子图。对于(2)，项目旨在基于模板模式识别空间和时间上的威胁。对于(3)，期望目标既是从微观到宏观尺度的行动的预测能力，也是从地方到区域到全国范围内评估预防措施的潜在影响与成本的工具。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

This project develops new mathematical algorithms and models involving knowledge graphs. A knowledge graph represents what is known about a subject in the form of labeled nodes and edges. More than simply labeled data, knowledge graphs organize data according to high-level meanings and assign globally unique identification to each node in the graph to match real-world entities. Much work on knowledge graphs treats databases and queries. In contrast, in the context of threat detection, this project focuses on algorithms that identify latent information in the graph and predictive models associated with data on the graph. The project will involve a combination of mathematical methods for subgraph isomorphism detection, time series analysis, agent-based and multiscale modeling, and pattern recognition. The project will train a postdoctoral scholar, PhD student, and six undergraduate researchers through involvement in the research.This project brings together several different focused problems with large, multimodal, complex datasets. The data is organized into a knowledge graph in which additional information is added and absorbed as it becomes available. This project considers three types of knowledge graphs each for different applications: (1) knowledge graphs constructed from complex multi-part narratives; (2) knowledge graphs constructed from heterogeneous online content; and (3) knowledge graphs associated with large-scale human interaction dynamics such as a global pandemic. For (1), algorithms will be designed to identify important causal subgraphs. For (2), the project aims to identify threats in space and time based on templated patterns. For (3), desired goals are both a predictive ability for actions from a micro to macro scale along with tools to assess potential impact versus cost of preventative measures, from local to regional to country-wide scale.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.