网络数据建模、分析与应用研究综述  

Network Data Modeling,Analysis and Application Review

作　　者：任怡萌 陶春柏 朱雪宁 Yimeng Ren;Chunbai Tao;Xuening Zhu(School of Data Science,Fudan University,Shanghai 200433;MOE Laboratory for National Development and Intelligent Governance,Fudan University,Shanghai 200433)

机构地区：[1]复旦大学大数据学院,上海200433 [2]复旦大学国家发展与智能治理综合实验室,上海200433

出　　处：《中国科学基金》2024年第5期773-783,共11页Bulletin of National Natural Science Foundation of China

摘　　要：互联网、大数据、人工智能等信息技术催生海量数据,网络数据作为重要数据形式,具有极高的挖掘潜力与分析价值。本文首先回顾了经典网络数据建模方法及相关理论性质研究,其次综述了上述方法在金融风险、宏观经济、商业营销、社会民生方面的具体应用。在此基础上,本文针对大数据背景下复杂网络数据异质性、非线性、高维度、大规模的特征,以及众多场景中的具体应用需求,总结现有研究存在的不足,阐述了网络数据建模在理论方法与实证分析中面临的挑战。最后,基于网络数据的新特点与新需求,给出现实场景中的网络数据分析在理论建模与应用研究方面的建议。The advent of information technologies such as the Internet,big data,and artificial intelligence has generated massive data.Network data,as a crucial form of data,has high mining potential and analysis value.This paper begins by reviewing classical methods for network data modeling and related theoretical properties.Subsequently,it surveys the specific applications of these methods in financial risk,macroeconomics,business marketing,and societal well-being.Furthermore,considering the heterogeneity,nonlinearity,high dimensionality,and large-scale features of network data,this paper identifies the shortcomings of the current research and outlines challenges faced in theoretical methods and empirical analysis of network data modeling under the background of massive data.Finally,based on the novel characteristics and demands of network data,the paper provides recommendations for theoretical modeling and applied research in analyzing network data in real scenarios.

关 键 词：复杂网络数据 网络自回归 异质性结构 非线性模型 高维数据分析 大规模网络 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论] O157.5[自动化与计算机技术—计算机科学与技术]