RAPID: COVIDGeoGraph – A Geographically Integrated Cross-Domain Knowledge Graph for Studying Regional Disruptions

COVID-19快速研究项目将开发一个地理集成知识图谱，以支持数据科学家、行业决策者和政府采取针对特定地区的措施，重新开放该国。为了实现这一目标，该项目将结合来自交通、社会距离措施、人口和环境因素以及经济影响等各种主题的数据。知识图谱是情境化技术。它们通过提供来自邻近学科的可操作的见解，使用户能够对复杂的社会和科学问题获得更全面的理解。例如，在做出有关当地经济和粮食系统的决策时，需要了解经常变化的社会距离措施、交通管制和限制（包括邻近地区）、人口因素和康复公民的百分比。该项目将与行业合作伙伴共同努力，了解如何在其预测模型中利用图形化知识。最后，该项目将延伸到其他知识图谱，共同形成与COVID-19相关的开放知识网络。在技术方面，我们将利用一堆开源技术和国际标准来开发一个高度集成的基于关联数据的知识图谱，该图谱结合了不同地理尺度和类型的地方的跨领域数据。我们将利用机器学习技术来学习图嵌入，以预测图中的链接，以及通过使用城市或县等地方作为集成点来与其他图对齐。我们还将提供与更广泛的schema.org协作的功能。我们特别感兴趣的是研究如何表示和对齐目前在不同地理尺度和总量上报告的COVID-19数据，从而促进互操作性和打破数据孤岛。我们将直接与我们的行业合作伙伴合作，从我们和外部图形中获得工程（空间明确的）特征。这些特征将被整合到工业下游预测模型中，特别关注粮食系统。最后，我们将利用合作伙伴的专业知识创建可视化数据仪表板，以更好地传达我们的发现。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 COVID-19 RAPID research program will develop a geographically integrated knowledge graph to support data scientists and decision-makers in industry and the government in taking region-specific steps towards reopening the country. Towards this goal, the project will combine data from themes as diverse as transportation, social distancing measures, demographic and environmental factors, as well as economic impacts. Knowledge graphs are contextualization technologies. They enable their users to gain a more holistic understanding of complex social and scientific questions by providing actionable insights from neighboring disciplines. For instance, making decisions about local economies and food systems require an understanding of the frequently changing social distancing measures, traffic control and restrictions including neighboring regions, demographic factors, and the percentage of recovered citizens. The project will work together with industry partners to understand how graphed knowledge can be utilized in their forecasting models. Finally, the project will reach out to other knowledge graphs to jointly form an open knowledge network related to COVID-19. More technically, we will utilize a stack of open source technologies and international standards to develop a highly integrated Linked Data-based knowledge graph that combines cross-domain data across different geographic scales and types of places. We will utilize machine learning technologies to learn graph embeddings for predicting links within our graph as well as alignments to other graphs by using places such as cities or counties as points of integration. We will also provide the functionality to collaborate with the broader schema.org effort. We are particularly interested in studying how to represent and align COVID-19 data that is currently being reported at different geographic scales and aggregates, thereby fostering interoperability and breaking up data silos. We will work directly with our industry partners on engineering (spatially-explicit) features from our and external graphs. These features will be integrated into industry downstream forecasting models with a particular focus on food systems. Finally, We will utilize the expertise of our partners in creating visual data dashboards to better communicate our findings.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.