基于可泛化模型的复杂商务场景数据分析  

Generalizable Analysis Methods for Complex Business Data

作　　者：史颖欢[1] 郭金涛 李泽昆 祁磊 高阳[1] Yinghuan Shi;Jintao Guo;Zekun Li;Lei Qi;Yang Gao(State Key Laboratory for Novel Software Technology,Nanjing University,Nanjing 210023;School of Computer Science and Engineering,Southeast University,Nanjing 210096)

机构地区：[1]南京大学计算机软件新技术全国重点实验室,南京210023 [2]东南大学计算机科学与工程学院,南京210096

出　　处：《中国科学基金》2024年第5期762-772,共11页Bulletin of National Natural Science Foundation of China

基　　金：国家自然科学基金项目(62222604,62206052,62192783)的资助。

摘　　要：随着全球经济合作与竞争形势的快速发展,复杂商务场景日益涌现。其中,数据的规模和复杂性也呈现显著增长,在带来可利用数据资源的同时,也为有效挖掘和分析潜在商务模式带来挑战。本文首先针对复杂商务场景下的数据特点进行分析,其次调研了商务分析方法的研究进展,包括:大规模商务数据集、多模态预训练与特征融合方法以及多场景下可泛化商务模型技术。此外,对复杂商务场景数据分析在未来的应用发展进行思考与总结,讨论了目前亟需解决的关键问题,包括:对更多模态的适应性、商务分析模型可解释性、动态环境下的分布变化鲁棒性等。进一步分析了复杂商务场景可泛化模型的研究现状和挑战,旨在为复杂商务分析领域的研究提供初步参考,实现复杂商务场景的全面智能化。With the rapid development of big data technology,business scenarios are rapidly evolving,witnessing a substantial increase in both the scale and intricacy of data.While it brings accessible data resources,it concurrently presents challenges in effectively excavating and analyzing business data.This paper first introduces the characteristics of data in complex business scenarios,then introduces the research advancements in business analysis methodologies,encompassing large-scale business datasets,multimodal pre-training,feature fusion and generalizable business models across diverse scenarios.The paper provides an outlook on the future research directions of data analysis in complex business scenarios,analyzing the key issues that need to be solved urgently,including adaptability to more modalities,interpretability of business analysis model,and generalizability in dynamic environments.The paper summaries the research progress and challenges encountered by generalizable models tailored for complex business scenarios,aiming to promote scientific research progress in complex business analysis and attaining comprehensive intelligence within complex business scenarios.

关 键 词：商务场景 可泛化数据分析 深度学习 多模态融合 

分 类 号：F49[经济管理—产业经济] TP311.13[自动化与计算机技术—计算机软件与理论]