Abstract:

The situation data in complex battlefield environment presents large-scale and fragmented features， the knowledge graph technology is applied to situation awareness to recognize operations from the perspective of integrity and relevance. The knowledge graph of operational action is constructed. The situation data such as operational action types， operational forces and attributes are organized in the form of graph to realize the correlation of operational knowledge. An operational action recognition model based on knowledge is designed. The model can fully utilize the attribute information and association information in the graph to enrich the vector representation of entities. By measuring the authenticity of triples， the operational action types and the relationships between combat forces can be recognized. Experimental results show that the proposed method can effectively recognize operations and help the commander to analyze the battlefield situation in depth.

Key words: knowledge association, knowledge graph, operational action recognition, representation learning, graph neural network, attention mechanism

摘要：

针对复杂战场环境中态势数据呈现出的大规模、碎片化特征，将知识图谱技术应用到态势感知工作中，从整体性和关联性的角度来判识作战行动。构建了作战行动知识图谱，将作战行动类型、作战力量以及属性等态势数据以图的形式进行组织，实现了作战行动知识的相互关联。设计了基于知识图谱的作战行动判识模型，该模型能够充分运用图谱中的属性信息和关联信息来得到更丰富的实体向量表示，通过衡量三元组的真实性来判识作战力量的作战行动类型以及作战力量之间的关系。实验表明，所提方法可以有效地对作战行动进行判识，有助于指挥人员深入地分析战场态势。

关键词: 知识互联, 知识图谱, 作战行动判识, 表示学习, 图神经网络, 注意力机制