柑橘病虫害知识图谱构建研究  

Study on the construction of knowledge graph for diseases and pests of citrus

作　　者：朱逸航 张小敏 周天奇 饶秀勤[1,2] ZHU Yihang;ZHANG Xiaomin;ZHOU Tianqi;RAO Xiuqin(College of Biosystem Engineering and Food Science,Zhejiang University,Hangzhou 310058,Zhejiang;Key Laboratory of Agricultural Products Processing Equipment,Ministry of Agriculture and Rural Affairs,Hangzhou 310058,Zhejiang;Zhejiang Kepler Technology Co.,Ltd.,Haining 314408,Zhejiang)

机构地区：[1]浙江大学生物系统工程与食品科学学院,浙江杭州310058 [2]农业农村部农产品产地处理装备重点实验室,浙江杭州310058 [3]浙江开浦科技有限公司,浙江海宁314408

出　　处：《浙江农业科学》2025年第3期687-691,共5页Journal of Zhejiang Agricultural Sciences

基　　金：国家重点研发计划项目(2023YFD1600300):乡村产业共性关键技术研发与集成应用。

摘　　要：在传统柑橘生产中,农户缺乏丰富的农业专业知识和技术。信息技术可以将柑橘病虫害防治知识数字化。然而,当前的知识管理技术在效率、可扩展性和适用性方面存在一些缺陷。知识图谱为知识管理提供了一种新的方法,它是一种更加灵活的知识管理方式。鉴于现有农业知识图谱对柑橘病虫害防治相关实体和关系的刻画不够详细,本研究对柑橘病虫害知识进行了实体类型和关系种类的定义,共划分出10种实体类别和10种实体关系,并利用大语言模型进行提取和去重,实现对16种常见缺陷的管理,使用Neo4j数据库对知识图谱进行了存储和可视化。最后着重探讨了柑橘病虫害知识图谱如何为精准病虫害信息查询、智能辅助诊断等下游任务提供底层技术支撑。In traditional cultivation of citrus,farmers lack abundant professional agricultural knowledge and technology.Information technology can digitize the knowledge of pest and disease control of citrus.However,current knowledge management technologies have some common deficiencies in terms of efficiency,scalability,and applicability.Knowledge graph provides a new approach to knowledge management,which is a more flexible method.Considering the insufficient detailed characterization of entities and relationships related to pest and disease control of citrus in existing agricultural knowledge graphs,this study defines entity types and relationship categories for pest and disease knowledge of citrus.A total of 10 entity categories and 10 entity relationships were identified,and a large language model was used for extraction and deduplication,addressing 16 common deficiencies.Finally,the knowledge graph was stored and visualized using the Neo4j database.The discussion focused on how knowledge graph for pest and disease of citrus can provide underlying technical support for downstream tasks such as precise pest and disease information inquiry and intelligent-assisted diagnosis.

关 键 词：柑橘病虫害 知识图谱 Neo4j 大语言模型 

分 类 号：S436.66[农业科学—农业昆虫与害虫防治]