摘要
随着企业智能化进程的加快,商业计算作为计算社会科学的重要分支,逐步从传统商业分析演进为融合大数据与人工智能的智能系统。然而,现有商业计算实践仍受制于多源数据的异构性与动态性、模型可解释性不足以及系统部署复杂度等问题,限制了其在复杂商业环境中的广泛应用。为应对上述挑战,本文提出智能时代的商业计算新范式,即以垂域大模型为驱动、以行业与领域为中心、真实业务场景为导向的商业计算体系。本文构建了“数据层—系统层—业务层—保障层”的四层研究框架,系统阐释了垂域大模型在商业计算中的技术路径与理论逻辑。该框架在有效整合多源异构数据、增强模型透明度与动态适应性及降低算力成本等方面展现出一定优势,能够提升商业计算系统在多元业务场景中的适应性与稳健性。进一步地,本文选取营销计算与组织计算两个典型应用场景,剖析了其关键科学问题,揭示了垂域大模型在应对外部市场动态与内部管理复杂性中的应用价值。最后,本文探讨了该新范式在人机协同机制、智能融合路径、信息更新策略及商业计算生态依赖等方面的研究难点与未来趋势,旨在为企业智能决策提供理论参考与方法支撑。
With the accelerating process of enterprise intelligence,business computing(i.e.,an important branch of computational social science)has evolved from traditional business analytics into intelligent systems integrating big data and artificial intelligence.However,current practices remain constrained by the heterogeneity and dynamism of multisource data,insufficient model interpretability,and the complexity of system deployment,which limit its broad application in complex business environments.To address these challenges,this paper proposes a domain-specific foundation modeldriven technical architecture as a potential breakthrough.It constructs a four-layer research framework encompassing the data layer,system layer,business layer,and governance layer,systematically outlining the evolution of business computing.The framework demonstrates advantages in effectively integrating multi-source heterogeneous data,enhancing model transparency and dynamic adaptability,and reducing computational costs,thereby improving the adaptability and robustness of business computing systems across diverse operational contexts.Building on this framework,the study examines two representative application contexts(i.e.,marketing computing and organizational computing)to systematically analyze their core scientific issues and showcase the value of domain-specific foundation models in addressing external market dynamics and internal management complexities.Finally,it outlines key future research directions for business computing,including human-AI collaboration mechanisms,intelligent integration pathways,information updating strategies,and ecosystem dependencies,aiming to provide theoretical guidance and methodological support for enterprise intelligent decision-making.
作者
杨帅
金佳
潘煜
Shuai Yang;Jia Jin;Yu Pan(Glorious Sun School of Business and Management,Donghua University,Shanghai 200051,China;Key Laboratory of Brain-Machine Intelligence for Information Behavior(Ministry of Education and Shanghai),Shanghai 201620,China;School of Business and Management,Shanghai International Studies University,Shanghai 201620,China)
出处
《中国科学基金》
北大核心
2025年第5期716-727,共12页
Bulletin of National Natural Science Foundation of China
基金
国家自然科学基金项目(72271166,72372023,71942003)的资助。
关键词
商业计算
垂域大模型
企业智能化
营销计算
组织计算
business computing
domain-specific foundation models
enterprise intelligence
marketing computing
organizational computing
