基于数据中台的多源异构数据指标分析方法  

Multi-source heterogeneous data indicator analysis method based on data middle platform

作　　者：李增伟 冶秀兰 马燕 刘青 赵一斌 LI Zengwei;YE Xiulan;MA Yan;LIU Qing;ZHAO Yibin(State Grid Qinghai Information&Telecommunication Company Co.,Ltd.,Xining 810000,China;State Grid Info-Telecom Greate Power Science and Technology Co.,Ltd.,Fuzhou 350003,China)

机构地区：[1]国网青海省电力公司信息通信公司,青海西宁810000 [2]国网信通亿力科技有限责任公司,福建福州350003

出　　处：《粘接》2024年第12期113-115,119,共4页Adhesion

摘　　要：针对信息技术发展中数据分类技术较落后的问题,研究设计了一个基于数据中台技术的企业财务数据多源集成系统。利用深度学习模型算法对中台内多源业务数据进行特征收集分析,将分析后的资源特征达到分类的目的。在中台系统内加入数据索引模型,来实现对分类数据的准确搜索。基于DTW优化索引技术,设数据特征阈值提升数据源搜索与索引的效率、速度及准确度。通过试验,该系统技术核算的数据,误差率在可接受范围内,该研究为其他技术研究奠定基础。In order to solve the problem of backward data classification technology in the development of information technology,a multi-source integration system of enterprise financial data based on data middle platform technology was designed.The deep learning model algorithm was used to collect and analyze the features of multi-source business data in the middle platform,and the analyzed resource characteristics were classified for the purpose of classification.A data index model was added to the middle office system to achieve accurate search of classified data.Based on the DTW optimized indexing technology,data feature thresholds were set to improve the efficiency,speed and accuracy of data source search and indexing.Through experiments,the data error rate of the technical accounting of the system was within an acceptable range,and the study lays the foundation for other technical studies.

关 键 词：中台技术 深度学习算法 多源异构 数据分类 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]