| 摘要: |
| 针对网络电磁空间信号类数据语义获取复杂、难以深度开发与组织利用的问题，本文提出了一种新的语义组织方式。提出了一种基于特征和行为两级数据类型的图谱建模方法。首先，根据业务场景的需求，明确需要采集与处理的数据类型，并按照特征级与行为级分别完成结构化、非结构化的存储。然后，基于存储内容的关键特征词完成上下文关联数据的抽取，并采用ALBERT-BiLSTM-CRF模型识别数据中的关键实体和关系。最后，通过关联数据知识融合的方法来构建语义增强的知识图谱。本方法能够有效地解决电磁空间数据语义转化为清晰、准确的响应序列输出问题，并通过雷达应用场景分析了本方法的可行性。 |
| 关键词: 电磁空间数据 特征建模 语义解析 知识图谱 |
| DOI： |
| 分类号:TP391 |
| 基金项目:乾元国家实验室基金项目 |
|
| [1]A Semantic Organization Method for Electromagnetic Space Data |
|
|
|
|
| Abstract: |
| Aimed at the challenges of complex semantic acquisition and difficulties in deep development and organized utilization of network electromagnetic space signal data, this paper proposes a novel semantic organization approach. A graph-based modeling method utilizing feature-behavior two-level data typology is introduced. First, based on the requirements of the business scenario, the data types to be collected and processed are defined, with structured and unstructured storage implemented for feature-level and behavior-level data respectively. Subsequently, contextual correlation data extraction is performed based on key feature terms of the stored content, where the ALBERT-BiLSTM-CRF model is employed to identify key entities and relation within the data. Finally, a semantically enhanced knowledge graph is constructed through correlated data knowledge fusion. This method effectively resolves the problem of transforming electromagnetic space data semantics into clear and actionable response sequence outputs, with analysis of Radar applications demonstrating its feasibility. |
| Key words: Electromagnetic Space Data Feature Modeling Semantic Parsing key word Knowledge Graph |