Abstract The complexity, variety and volume of urban datasets have soared in the past decade. By utilising these datasets, urban planners and researchers have built a wide range of evidence-based methods and analytical tools for planning and decision-making purposes. However, the data heterogeneity remains one of the key problems in this process. Building generic urban analytics tools adaptable to diverse data formats remains largely unresolved in practice. This work proposes an innovative system called Urban Data Analytics Infrastructure (UADI) to tackle these challenges by leveraging on the advantages of ontology technologies. The proposed technique implements a bi-level mapping approach to consolidate heterogeneous datasets into a uniformed structure. This is presented in ontology schemas and hence offers a new means for developing generic tools for urban analytics. By applying bi-level mapping between data and ontology, the datasets are semantically enriched. This strategy allows data harmonisation, thus, heterogeneity problems could be mitigated. When building an analytics tool, researchers can simply declare the input data type as a specific concept of ontology and then follow the ontology schema to implement the code. The developed tool can be registered into the UADI system and then can work with any data mapped to the concept. The core components of the system include data registration, data enrichment, ontology management, translation engine, tool development and tool management. These are elaborately designed and developed to meet the design goals. The system usability and performance are validated by building a series of ISO 37120 indicators (for city services and quality of life) within the UADI.

中文翻译：

---

基于语义的城市分析数据基础设施的设计与实践

摘要 在过去十年中，城市数据集的复杂性、多样性和数量激增。通过利用这些数据集，城市规划者和研究人员为规划和决策目的建立了广泛的循证方法和分析工具。然而，数据异质性仍然是这一过程中的关键问题之一。构建适用于不同数据格式的通用城市分析工具在实践中在很大程度上仍未解决。这项工作提出了一种称为城市数据分析基础设施 (UADI) 的创新系统，以利用本体技术的优势来应对这些挑战。所提出的技术实现了一种双层映射方法，以将异构数据集整合到一个统一的结构中。这在本体模式中呈现，因此为开发用于城市分析的通用工具提供了一种新方法。通过在数据和本体之间应用双层映射，数据集在语义上得到了丰富。该策略允许数据协调，因此可以减轻异质性问题。在构建分析工具时，研究人员可以简单地将输入数据类型声明为本体的特定概念，然后按照本体模式来实现代码。开发的工具可以注册到 UADI 系统中，然后可以处理映射到概念的任何数据。该系统的核心组件包括数据注册、数据丰富、本体管理、翻译引擎、工具开发和工具管理。这些都是精心设计和开发的，以满足设计目标。