图书馆杂志

图书馆杂志 ›› 2025, Vol. 44 ›› Issue (414): 13-22.

• “国家科研论文和科技信息高端交流平台建设”专题 • 上一篇    下一篇

国外高端交流平台生成式人工智能融合模式研究

徐宏宇(上海图书馆)   

  • 出版日期:2025-10-15 发布日期:2025-11-06
  • 作者简介:

    徐宏宇 上海图书馆( 上海科学技术情报研究所),研究员。研究方向: 竞争情报与科技决策。E-mailhyxu@libnet. sh. cn 上海 200031

The Integration Models of Generative Artificial Intelligence inInternational Advanced Platforms for S&T Information and Communication

Xu Hongyu (Shanghai Library)   

  • Online:2025-10-15 Published:2025-11-06
  • About author:Xu Hongyu Shanghai Library

摘要:

本文围绕国外高端交流平台与生成式人工智能(Generative AI) 的融合实践展开系统研究,梳理并对比传统平台与新兴平台的AI 应用路径及其发展特征。首先,文章区分了传统与新兴高端交流平台的结构生态:前者包括大型学术出版集团、专业信息服务商与公共机构,具有内容权威性强、服务链条完整等特点;后者则以预印本平台、开放数据集与非官方文献平台为主,强调开放共享与协作创新。在此基础上,文章总结出传统平台的三类典型AI 融合模式:自主研发垂直AI、集成外部大模型战略合作、平台内容授权合作,体现了其中心化管理与商业化导向的特征。针对新兴平台,文章则提炼出数据与接口开放、协作众包、自研工具嵌入三种AI 融合路径,突出其生态开放性和开发者驱动特征。最后,通过对标研究,文章指出两类平台的AI 融合呈现“中心化闭环” 与“ 生态化开放”两种不同逻辑,为高端交流平台的新发展提供了实践启示和理论参考。

关键词: 高端交流平台&emsp, 生成式人工智能&emsp, AI 融合路径&emsp, 开放科学&emsp, 学术出版转型

Abstract:

This paper presents a systematic study of the integration of generative AI into advancedplatforms for S&T information and communication in the international context. It analyzes and comparesthe AI application pathways and development characteristics of both traditional and emerging platforms.The paper first distinguishes between the structural ecologies of the two types of platforms traditionalplatforms—including major academic publishers professional information service providers and publicinstitutions—are characterized by strong content authority and complete service chains emergingplatforms such as preprint servers open data repositories and unofficial literature platforms emphasizeopenness sharing and collaborative innovation. Based on this distinction the paper identifies threetypical AI integration models among traditional platforms vertically self-developed AI strategicpartnerships with large external models and content licensing collaborations highlighting a trend towardcentralized management and commercial orientation. For emerging platforms it outlines three AIintegration pathways data and interface openness collaborative crowdsourcing and embedded selfdevelopedtools which reflect an ecosystem-driven and developer-oriented logic. Finally throughcomparative analysis the paper concludes that the two types of platforms exhibit distinct integrationlogics the centralized and closed-loop model versus the ecosystem-oriented and open model. Thesefindings provide both practical insights and theoretical references for the future development of advancedcommunication platforms.

Key words:

Advanced platforms for S&, T information and communication Generative artificial intelligenceAI integration pathways Open science Academic publishing transformation