建设管理中合同文档关键信息智能抽取技术研究及应用  

Research and Application of Intelligent Extraction Technology of Key Information of Contract Document in Construction Management

作　　者：李晓波 刘勇[1] LI Xiaobo;LIU Yong(College of Computer and Information Science,China Three Gorges University,Yichang 443002)

机构地区：[1]三峡大学计算机与信息学院,湖北宜昌443002

出　　处：《长江信息通信》2025年第2期158-160,175,共4页Changjiang Information & Communications

摘　　要：建设管理系统中合同电子文档需人工录入合同关键信息,该方式不仅工作效率低、劳动强度大,而且通常会出现人为输入错误,存在较大的管理风险。因此,文章提出了一种建设管理中合同文档的关键信息的智能抽取方法,该方法首先使用基于规则的文本片段截取算法截取含有合同录入信息的文本片段,再通过基于BERT-BiLSTM-CRF模型地中文命名实体识别技术提取出合同文档需要录入的关键信息。统计数据表明,该智能抽取方法能够比较准确的抽取出合同文档中信息,满足建设管理中合同录入需要。The electronic document of contract in the construction management system needs manual input of key information of contract,which not only has low work efficiency and high labor intensity,but also usually causes human input errors,which has a large management risk.Therefore,this paper proposes an intelligent extraction method for key information of contract documents in construction management.The method first uses rule-based text fragment extraction algorithm to extract text fragments containing contract input information,and then extracts key information of contract documents through named entity recognition technology based on BERT-BiLSTM-CRF model.The statistical data show that the intelligent extraction method can extract the information from the contract document more accurately,and meet the needs of contract input in construction management.

关 键 词：合同录入 规则抽取 命名实体识别 深度学习 机器学习 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]