Basic knowledge graph of military unmanned systems

To explore the technical path for constructing knowledge graphs in the military field based on open evaluation tasks, the team collaborated with the China Conference on Knowledge Graph and Semantic Computing(CCKS). In 2021 and 2022, they successively released the evaluation tasks for constructing knowledge graphs of military unmanned systems through the CCKS evaluation task section. Namely, "CCKS2021 - Construction of Vertical Domain Knowledge Graph for Military Unmanned Aerial Vehicle Systems" and "CCKS2022 - Construction of Knowledge Graph for Foreign Military Unmanned Aerial Vehicle Systems". Driven by the design of the standard ontology, task definition and evaluation indicators of unmanned systems, encyclopedic and authoritative datasets in the field of military unmanned systems were collected and selected. Through the approach of "release + evaluation + crowdsourcing + verification", a high-quality knowledge graph of military unmanned systems was constructed.In 2023, to support the implementation of the "CCKS2023 - Reasoning and Question-Answering of Knowledge Graph of Foreign Military Unmanned Systems" task, the team extracted the "subgraphs" related to the basic knowledge of unmanned systems from the graph and named it "Basic Knowledge Graph of Military Unmanned Systems". This graph consists of three csv data files: entities, relations, and attributes, namely entiti.csv, relation.csv, and attribute.csv. The data volume is approximately 8,000 entities, 14,000 relations, and 36,000 attributes, forming a relatively complete basic knowledge network for military unmanned systems.

为探索基于开放评测任务构建军事领域知识图谱的技术路径，团队与中国知识图谱与语义计算大会（China Conference on Knowledge Graph and Semantic Computing，CCKS）展开合作。2021年与2022年，团队先后通过CCKS评测任务板块发布了军事无人系统知识图谱构建相关评测任务，分别为"CCKS2021——军事无人机系统垂直领域知识图谱构建"与"CCKS2022——外军无人机系统知识图谱构建"。依托无人系统标准本体、任务定义与评测指标的设计框架，团队收集并筛选了军事无人系统领域的权威百科类数据集。通过"发布+评测+众包+验证"的技术路径，最终构建出高质量的军事无人系统知识图谱。2023年，为支撑"CCKS2023——外军无人系统知识图谱推理与问答"任务的落地实施，团队从该图谱中抽取与无人系统基础知识相关的子图，并将其命名为"军事无人系统基础知识图谱"。该图谱包含三个CSV格式数据文件：实体表、关系表与属性表，即entiti.csv、relation.csv与attribute.csv。其数据规模约为8000个实体、14000条关系以及36000条属性，构建起较为完整的军事无人系统基础知识网络。