盾构施工知识自动抽取与建模  

Automatic Extraction and Modeling of Shield Construction Knowledge

作　　者：刘江波 徐宏[3] 王浩茜 徐晟 LIU Jiangbo;XU Hong;WANG Haoxi;XU Sheng(School of Economics and Management,Chang’an University,Xi’an 710064,China;Project of Digital Construction and Management for Major Transportation Infrastructure along“Belt and Road”Initiative in Shaanxi Province,Chang’an University,Xi’an 710064,China;China Railway First Bureau Group Co,Ltd,Xi’an 710054,China)

机构地区：[1]长安大学经济与管理学院,陕西西安710064 [2]长安大学陕西省“一带一路”重大交通基础设施数字化建设与管理学科创新引智基地,陕西西安710064 [3]中铁一局集团有限公司,陕西西安710054

出　　处：《土木工程与管理学报》2023年第2期130-137,共8页Journal of Civil Engineering and Management

基　　金：国家重点研发计划(2021YFB2601000);国家自然科学基金(51708039)。

摘　　要：建立领域知识图谱有利于解决盾构施工知识分散,检索效率低下的问题,但现有知识抽取方法较为复杂,且较多地依赖于人工参与,因此,提出一种以规范文本为对象的知识自动抽取方法。该方法在条文预处理和条文分类的前提下,以现有自然语言处理平台预训练模型提供的语义角色标注(SRL)结果为基础,通过后优化处理和模式匹配实现知识抽取,形成不同类型条文关键知识要素的结构化整理。最后,以GB 50446—2017《盾构法隧道施工及验收规范》为例,进行知识抽取并建立了盾构施工知识图谱。结果表明,该方法能在一定程度上降低中文规范文本的抽取难度,为领域知识图谱的建立提供有效的数据支撑,从而促进盾构施工知识的高效积累、共享和重用。The establishment of domain knowledge graph is beneficial in addressing the problems of scattered knowledge and inefficient retrieval of shield construction.Existing knowledge extraction methods are more complicated and rely on manual engagement.This paper proposes an automatic knowledge extraction method based on standard normative text.Under the premise of text pre-processing and clauses classification,the method achieves knowledge extraction through post-optimization processing and pattern matching based on the Semantic Role Labeling(SRL)results provided by the pre-training model of the existing natural language processing platform,and then forms a structured organization of key knowledge elements of different types of clauses.Finally,take the GB 50446—2017“Code for Construction and Acceptance of Shield Tunnelling Method”as an example to extract knowledge and construct a shield construction knowledge graph.The results show that this method can reduce the difficulties of extracting Chinese specification text to some extent and provide effective data support for the creation of domain knowledge graph,promote the efficient accumulation,sharing and reuse of shield construction knowledge.

关 键 词：知识抽取 语义角色标注 本体 知识图谱 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]