大数据驱动的复杂系统管理情景建模:技术与流程  

Modeling of Complex System Management Scenarios Driven by Big Data: Techniques and Processes

作　　者：王海燕[1] 盛昭瀚[2] Wang Haiyan;Sheng Zhaohan(School of Economics and Management,Southeast University,Nanjing 211189,China;School of Management and Engineering,Nanjing University,Nanjing 210093,China)

机构地区：[1]东南大学经济管理学院,江苏南京211189 [2]南京大学工程管理学院,江苏南京210093

出　　处：《中国管理科学》2025年第1期22-33,共12页Chinese Journal of Management Science

基　　金：国家社会科学基金重大项目(23&ZD049);南京大学中国特色哲学社会科学自主知识体系建构“引领工程”重大研究专项(2024300573)。

摘　　要：为了破解复杂系统管理活动中的复杂整体性,基于管理活动在过去、现在和将来的任何一个时间点上都有情景的生成与演化这一特征,以情景作为抓手,通过对复杂系统管理情景建模再现复杂系统过去的情景、重构复杂系统现在的情景、预测复杂系统将来的情景,为复杂系统管理活动提供实践指导。基于大数据和情景在学理上的同一性,分析了大数据驱动复杂系统管理情景建模的概念与内涵,明确了大数据驱动的复杂系统管理情景建模就是大数据驱动实现复杂系统管理情景模型化这一理念。从观测或实验手段采集或记录获得的大数据构成的情景数据集入手进行逆向建模,提出了这一建模过程中需要用到的情景结构化、情景数据的大数据化、核情景提炼、情景耕耘、情景校核和联邦式建模6大类关键技术,设计了探索性分析-概念情景构建-数据情景构建-核情景构建-计算情景构建-情景生成与校核这一具体建模流程,为复杂系统管理情景建模的实施和应用提供依据。In order to break through the complexity integrity of complex system management activities, based on the characteristic that management activities generate and evolve scenarios at any time point in the past, present, and future, using scenarios as a starting point, modeling complex system management scenarios to reproducing the past scenarios, reconstructing the current scenarios, and predicting the future scenarios of complex systems are used to provide practical guidance for complex system management activities. Based on the theoretical similarity between big data and scenarios, the concept and connotation of big data-driven complex system management scenario modeling are analyzed and it is clarified that big data-driven complex system management scenario modeling is the concept of big data-driven implementation of complex system management scenario modeling. Starting from the scenario dataset composed of big data collected or recorded through observation or experimental methods, reverse modeling is carried out. Six key technologies are proposed for this modeling process, including scenario structuring, big data transformation of scenario data, kernel scenario extraction, scenario cultivation, scenario verification, and federated modeling. A specific modeling process is designed, including exploratory analysis-conceptual scenario construction-data scenario construction-kernel scenario construction-computational scenario construction-scenario generation and verification, providing a basis for the implementation and application of scenario modeling in complex system management.

关 键 词：大数据驱动 复杂系统管理 情景建模 建模技术 建模流程 

分 类 号：C93[经济管理—管理学]