欢迎访问《农学学报》,

农学学报 ›› 2023, Vol. 13 ›› Issue (7): 80-86.doi: 10.11923/j.issn.2095-4050.cjas2022-0067

• 农业信息 农业气象 • 上一篇    下一篇

土壤肥力与农作物科学施肥管理融合知识图谱构建与可视化

张彩丽1(), 吴赛赛2, 李玮3, 王慧4, 陈磊1()   

  1. 1 安徽省农业科学院农业经济与信息研究所,合肥 230031
    2 中国农业科学院农业信息研究所,北京 100081
    3 安徽省农业科学院作物研究所,合肥 230031
    4 安徽省农业科学院土壤肥料研究所,合肥 230031
  • 收稿日期:2022-06-08 修回日期:2022-12-01 出版日期:2023-07-20 发布日期:2023-07-18
  • 通讯作者: 陈磊,男,1977年出生,安徽金寨人,副研究员,硕士,从事农业信息化研究。E-mail:chenleichina@163.com
  • 作者简介:

    张彩丽,女,1980年出生,安徽固镇人,助理研究员,博士,主要从事农业信息分析、农业知识图谱研究。E-mail:

  • 基金资助:
    国家自然科学基金项目“绿肥翻压下双季稻区稻田土壤磷素形态转化及机理研究”(41807106); 国家重点研发计划项目子课题“砂姜黑土丰产增效土壤培肥和绿色产品应用模式与示范”(2016YFD0300809-5); 安徽省农业科学院2021年定向委托类项目“基于知识图谱的安徽省土壤肥力知识问答系统”(2021YL053)

Knowledge Graph Construction and Visualization in Soil Fertility and Scientific Crop Fertilizer Management

ZHANG Caili1(), WU Saisai2, LI Wei3, WANG Hui4, CHEN Lei1()   

  1. 1 Agricultural Economy and Information Research Institute, Anhui Academy of Agricultural Sciences, Hefei 230031, Anhui, China
    2 Agricultural Information Institute, Chinese Academy of Agricultural Sciences, Beijing 100081, China
    3 Crop Research Institute, Anhui Academy of Agricultural Sciences, Hefei 230031, Anhui, China
    4 Soil and Fertilizer Research Institute, Anhui Academy of Agricultural Sciences, Hefei 230031, Anhui, China
  • Received:2022-06-08 Revised:2022-12-01 Online:2023-07-20 Published:2023-07-18

摘要:

为促进人工智能各项技术在农业生产中发挥作用,探索土壤肥力领域知识的组织、应用,采用人工构建的方法,以安徽省为例,使用Protégé本体库构建工具构建土壤肥力领域本体,再将土壤肥力相关的结构化、半结构化和非结构化数据通过纠错等操作后采取反距离加权插值,基于ERNIE-BiLSTM-CRF与PCNN-Attention深度学习模型实现命名实体识别和关系抽取任务,其后把得到的三元组数据保存至图数据库Neo4j中,成功构建土壤肥力可视化知识图谱。该研究可在本体构建、实体关系抽取模型以及知识图谱的可视化方面为其他农业知识图谱的构建提供参考。

关键词: 土壤肥力, 本体构建, 实体关系抽取, 深度学习

Abstract:

Aiming to improve the function of artificial intelligence in agriculture, and to search the knowledge organization and application in field of soil fertility, the soil fertility ontology of Anhui Province was built with manual construction method as well as the Protégé application software. Then the structured data, semi-structured data and unstructured data related to soil fertility in Anhui Province were first cleaned and then processed by inverse distance weighted (IDW) interpolation. Based on deep learning model of ERNIE-BiLSTM-CRF and PCNN-Attention, the tasks of named entity recognition and relationship extraction were realized, and all the triplet data were stored in Neo4j graph database. And the visual knowledge graph of soil fertility was successfully constructed. This study can provide reference for the construction of other agricultural knowledge graph in the aspects of ontology construction, entity relationship extraction model and graph visualization.

Key words: soil fertility, ontology construction, entity relationship extraction, deep learning