面向建筑施工安全规范的自动结构化方法  

The Automatic Structured Method of Construction Safety Rules

作　　者：吕莹[1] 张智博 苏义坤[1] 郑志哲 LV Ying;ZHANG Zhibo;SU Yikun;ZHENG Zhizhe(College of Civil Engineering and Transportation,Northeast Forestry University,Harbin 150040,China;China Geological Material Corporation,Beijing 100098,China)

机构地区：[1]东北林业大学土木与交通学院,黑龙江哈尔滨150040 [2]中国地质物资有限公司,北京100098

出　　处：《工程管理学报》2025年第1期121-127,共7页Journal of Engineering Management

基　　金：国家重点研发计划项目(2021YFF0602000).

摘　　要：为解决建筑施工安全规范条文分散、检索效率低的问题,提出一种基于语义依存分析的建筑施工安全规范条文的自动结构化方法。采用自顶向下的方式建立建筑施工安全本体模型框架,利用现有的自然语言处理平台LTP4.0对规范条文语句进行语义依存分析构建语义依存图,并借助正则表达式和关键字识别提取语义依存图中各节点语言单位的概念特征。在此基础上,结合语句中的依存关系和概念特征推导出各类规范条文的判断依据,实现规范条文的自动化抽取,从而构建知识图谱数据层。并通过应用分析验证自动化抽取的方法流程,较高的准确率表明此方法流程能在一定程度上降低规范条文的抽取难度,为开拓建筑施工安全规范条文集成和应用提供了新思路,有助于提升建筑施工安全管理的信息化水平。To address the issues of scattered knowledge and low retrieval efficiency of the construction safety standards,an au-tomatic structuring method for the construction safety standards is proposed.A top-down approach is used to establish an ontology model framework for building construction safety.The existing natural language processing platform,LTP4.0,analyzes the semantic dependencies of standards to construct a dependency graph.The conceptual features of each node in the graph are extracted using regular expressions and keyword recognition.Based on this,the judgment basis for various normative provisions is deduced by com-bining dependencies and conceptual features in the sentences,enabling the automatic extraction of standards and the construction of the knowledge graph data layer.Lastly,the method process is verified through case analysis,and the high accuracy demonstrates that the method can reduce the difficulty of extracting the standards to a certain extent,provides new ideas for the integration and appli-cation of construction safety standard knowledge,and helps to improve the information level of construction safety management.

关 键 词：建筑施工安全 规范条文 结构化 语义依存分析 知识图谱 

分 类 号：TU714[建筑科学—建筑技术科学]