摘要

近些年来，国内隐患排查形式由传统的现场排查整改向信息化管理平台搭建的转变，北京、天津、青岛、武汉等多个城市已经开始隐患排查系统的搭建，实现地铁施工安全隐患排查线上、实时、动态管理。隐患排查标准是隐患排查系统中排查人进行隐患识别的重要参照依据，而现有的隐患排查标准中各隐患条文数量繁多，但都是独立存在的，缺乏对隐患概念的关联性的梳理，并且多采用文档的方式，不利于计算机的处理与分析，使得隐患信息在计算机中的应用存在困难。本文的目的在于提出探究隐患排查标准中核心概念的语义关联，以语义网络建立不同隐患内容间的关联，并将该方案应用于隐患内容的检索，以增强隐患知识的检索效率。

地铁施工安全隐性知识建模与应用的研究主要包括隐患排查标准语义网络构建与基于语义网络的规范条文检索在隐患排查系统中的应用两部分。以地铁施工起重吊装类隐患知识点为例，首先结合How Net 的构建思想与技术，分析归纳起重吊装类隐患知识的5 类基础语义节点以及23 类语义关系，利用protege构建了该类知识的语义网络模型，实现知识概念与关系的图示展示。在此基础上，通过建立查询文本的分词规则、语义标注规则以及知识间基于语义网络的相似度计算方法，设计隐患排查系统中隐患排查标准的检索方案，用户利用隐患排查系统进行隐患信息登记时，能够快速从大量隐患排查标准项中检索到所需隐患知识，从而提高隐患排查的效率。

关键词：地铁施工安全隐患知识建模语义网络相似度

Abstract

In recent years, means ofthe potential safety hazards management changes from woke site checking toinformation management platform building in China, which have been developed inBeijing, Tianjin, Guangzhou, Wuhan and some other cities, in order to supportsthe online, real-time and dynamic management of potential safety hazards. Constructionsafety management regulations plays an important in assuring safety. For onething ,there are various of regulations, in which each of the item is independentand each of the concept lack of relevance, For other thing, the regulations arepaper-based and not conducive to computer analysis. As a result ,it is difficultfor users to make use of these regulations in intelligent systems. The purpose ofthis paper is to construction the semantic work of different concepts in safetyhazards management, and explore the semantic relations. the semantic network isproposed to apply in potential safety hazards management system to improve the regulationretrieval efficiency.

Two methods are applied inknowledge modeling and application in subway construction safety hazards：the use of semanticnetwork to construct the knowledge structure between concepts, and the semanticretrieval method of query text and text-based specifications. Hoisting hazardsare taken as an example in this paper. At first,How Net technology and semanticnetwork construction methods are combined, five semantic types and twenty-threekinds of semantic relations are analyzed and summarized,semantic network modelis established by protege,which illustrates semantic concepts and relationshipsclearly. On this basis, the rules of query text segmentation and semanticannotation of are established, the similarity between different kind ofknowledge are calculated,the system visualization scheme applied to SafetyHazards Inspection System is proposed and analyzed,which may achieve the functionthat query text can be used to retrieve specification, and site pictures becomemedium of tacit knowledge visualization.

Keywords:subway construction   potential safetyhazards   knowledge modeling

semantic network   similarity