摘要
首先针对工程文档的格式特点,提出支持多种来源、多种格式文档的统一文本提取和语义分析方法,获取文档中蕴含的关键信息。然后提出一种基于改进TF-IDF指标和主成分分析(PCA)的文档指纹算法和文档知识网络建立算法。最后基于建筑工业互联网平台开发相应的功能模块;并将100多个大型项目的40万个文档进行分析和组织,按主题词和关键短语的内在联系,形成跨项目可视化可交互的知识网络。
According to the format characteristics of engineering documents,this paper proposes a unified text extraction and semantic analysis method supporting multi-source and multi format documents.After that,proposes a document fingerprint algorithm and a document knowledge network building algorithm based on improved TF-IDF index and principal component analysis(PCA).Finally,developes the software modules on the building industry internet platform for analyzing and organizing 400 thousand documents of more than 100 large projects.
作者
彭阳
左锋
余芳强
张铭
PENG Yang;ZUO Feng;YU Fangqiang;ZHANG Ming(Shanghai Construction No.4(Group)Co.,Ltd,Shanghai 201103,China;Xinhua Hospital Affiliated to Shanghai Jiaotong University School of Medicine,Shanghai 200092,China)
出处
《建筑经济》
北大核心
2021年第S01期411-414,共4页
Construction Economy
基金
上海市工业互联网创新发展专项(2019-GYHLW-01002)
关键词
工程文档
知识网络
工业互联网
语义分析
文档指纹
engineering documents
knowledge network
industry internet
semantic analysis
document fingerprint
