摘要
为了提高海关实验室检索效率和语义理解能力,构建以知识图谱为核心的智能检索体系。系统围绕“样品—项目—方法—设备—标准”等核心实体设计本体建模结构,融合异构数据建立统一语义模型。架构采用三层解耦设计,集成基于BERT-Attention的意图识别、DSL语义解析、图谱调度与结果融合等模块。典型任务验证表明,该体系在查询响应时间、Top-5命中率和复杂语义适配能力方面均优于传统方式,显著提升多源数据的语义组织与检索效率。
In order to improve the retrieval efficiency and semantic understanding capabilities of customs laboratories,this study constructs an intelligent retrieval system centered on knowledge graphs.The system designs an ontology modeling structure around core entities such as“sample-item-method-equipment-standard,”integrating heterogeneous data to establish a unified semantic model.The architecture adopts a three-layer decoupled design,integrating modules including BERT-Attention-based intent recognition,DSL semantic parsing,graph scheduling,and result fusion.Validation on typical tasks demonstrates that this system outperforms traditional approaches in query response time,Top-5 hit rate,and complex semantic adaptation capabilities,significantly enhancing the semantic organization and retrieval efficiency of multi-source data.
作者
赵波
郭天慧
高超
ZHAO Bo;GUO Tianhui;GAO Chao(Qingdao Customs Technology Center,Qingdao Shandong 266114,China;Certification and Accreditation Technology Research Center,State Administration for Market Regulation,Beijing 100088,China)
出处
《口岸非传统安全学刊》
2025年第6期72-76,共5页
JOURNAL OF NON-TRADITIONAL BORDER SECURITY SCIENCE AND TECHNOLOGY
基金
海关总署科技项目(2025HK252)
海关总署科技项目(2025HK002)。
关键词
海关实验室
知识图谱
语义检索
customs laboratory
knowledge graph
semantic retrieval
