针对现有地理知识图谱构建流程复杂、扩展性和动态性不足的问题，提出了一种基于动态本体的跨维地理知识图谱构建系统。系统通过引入动态本体技术，支持本体模型的实时更新和扩展，结合RESTful API与WebSocket技术，确保本体操作的高效性和实时性。在数据处理方面，针对结构化、半结构化和非结构化地理数据，分别采用Pandas、模板匹配和ALBERT模型进行数据清洗、实体识别与关系抽取，生成RDF三元组存储于MongoDB中。系统利用Canvas和Google Earth API，实现地理知识的二维画布与三维地球的可视化展示，增强了知识图谱的交互性与直观性。同时，通过图计算算法对图谱数据进行分析，为地理信息的深层次挖掘和辅助决策提供支持。实验表明，系统在动态本体管理、数据抽取与跨维图谱展示方面具有高效性和灵活性，适用于多源异构地理数据的通用管理与分析需求，为地理知识图谱的构建与应用提供了一种新思路。

In view of the problems of complex construction process, insufficient scalability and dynamism of existing geographical knowledge graph, we proposed a cross-dimensional geographical knowledge graph construction system based on dynamic ontology. By introducing dynamic ontology technology, the system supports real-time update and expansion of ontology model, and combines RESTful API and WebSocket technology to ensure the efficiency and real-time performance of ontology operation. In terms of data processing, Pandas, template matching and ALBERT model are used for data cleaning, entity recognition and relationship extraction for structured, semi-structured and unstructured geographical data,respectively, and RDF triples are generated and stored in MongoDB. The system uses Canvas and Google Earth API to realize the visualization of 2D canvas and 3D Earth of geographical knowledge, which enhances the interactivity and intuitiveness of knowledge graph. At the same time,the graph data is analyzed by graph computing algorithm to provide support for deep mining and decision-making support of geographical information. Experiments showed that the system was efficient and flexible in dynamic ontology management, data extraction and cross-dimensional graph display, which was suitable for the general management and analysis needs of multi-source heterogeneous geographical data, and could provide a new idea for the construction and application of geographical knowledge graph.