水利学报2024，Vol.55Issue(9)：1071-1083,1097,14.DOI:10.13243/j.cnki.slxb.20230800
基于本体和自然语言处理的土石坝险情知识图谱构建方法研究
An ontology and NLP-based knowledge graph construction method for dangers of earth-rock dam
摘要
Abstract
Due to factors such as suboptimal construction quality and severe environmental disasters,a range of dangerous situations,including landslides,cracks,and flooding,may arise during the operational and mainte-nance stages of earth-rock dams.At present,a large amount of data related to dangers of earth and rock dam is dispersed and exists in diverse forms,posing challenges for its conversion into experiential knowledge for efficient utilization and swift guidance in hazard mitigation.Consequently,a knowledge graph(KG)construction method is proposed in this study,integrating ontology and Natural Language Processing(NLP).To construct pattern and data layers of the graph,this method employs the top-down and bottom-up methods,respectively.The pattern layer centers on three primary concepts:risk types,risk causes,and risk disposal measures.A domain ontology library is established encompassing four facets of earth-rock dam structure,process,environment,and materi-als,and a conceptual framework of KG is built.The construction of the data layer encompasses operations such as data preprocessing,knowledge extraction,and semantic alignment.NLP is employed to process the text and for-mulate extraction rules based on corpus characteristics,facilitating the acquisition of specific knowledge content for the data layer.Ultimately,various types of instances and their relationships are stored in the form of triplets.The Neo4j platform is employed to actualize the visual representation and querying application of the KG within the dangers of earth-rock dam.This transformation of dispersed data in the domain into integrated knowledge offers essential technical and theoretical support for the safety management and hazard mitigation of earth-rock dams.关键词
土石坝险情/知识图谱/本体/自然语言处理Key words
dangers of earth-rock dam/knowledge graph/ontology/natural language processing分类
建筑与水利引用本文复制引用
张继勋,王虞清,焦修明,张玉贤..基于本体和自然语言处理的土石坝险情知识图谱构建方法研究[J].水利学报,2024,55(9):1071-1083,1097,14.基金项目
浙江省水利河口研究院院长科学基金项目(ZIHE21Z004) （ZIHE21Z004）
云南省重大科技专项计划项目(202102AF080001) （202102AF080001）
