中国科技期刊研究 ›› 2021, Vol. 32 ›› Issue (1): 65-74. doi: 10.11946/cjstp.201911220799

• 数字出版 • 上一篇    下一篇

人工智能辅助学术同行评议的应用及分类

张彤1(), 尹欢2, 苏磊3, 王静1, 夏道家1()   

  1. 1)《南京航空航天大学学报》《南京航空航天大学学报(英文版)》《数据采集与处理》编辑部,江苏省南京市秦淮区御道街29号 210016
    2)南京农业大学《园艺研究》编辑部,江苏省南京市玄武区卫岗1号 210095
    3)北京航空航天大学文化传媒集团《航空学报》编辑部,北京市海淀区学院路37号 100083
  • 收稿日期:2019-11-22 修回日期:2020-10-19 出版日期:2021-01-15 发布日期:2021-01-15
  • 通讯作者: 夏道家 E-mail:lee4@nuaa.edu.cn;xdjym@nuaa.edu.cn
  • 作者简介:张 彤(ORCID:0000-0001-9942-3879),博士研究生,编辑,E-mail: lee4@nuaa.edu.cn;|尹 欢,博士研究生,编辑部主任;|苏 磊,硕士,编辑;|王 静,博士,编辑。
  • 基金资助:
    中国科技期刊卓越行动计划选育高水平办刊人才子项目—青年人才支持项目(2020ZZ111042)

Application and classification of AI-assisted academic peer review

ZHANG Tong1(), YIN Huan2, SU Lei3, WANG Jing1, XIA Daojia1()   

  1. 1) Editorial Office of Journal of Nanjing University of Aeronautics and Astronautics, Transactions of Nanjing University of Aeronautics and Astronautics, and Journal of Data Acquisition and Processing, 29 Yudao Street, Qinhuai District, Nanjing 210016, China
    2) Editorial Office of Horticlulture Research, Nanjing Agricultural University, 1 Weigang, Xuanwu District, Nanjing 210095, China
    3) Editorial Office of Acta Aeronautica et Astronautica Sinica, Cultural Media Group Ltd., Beihang University, 37 Xueyuan Road, Haidian District, Beijing 100083, China
  • Received:2019-11-22 Revised:2020-10-19 Online:2021-01-15 Published:2021-01-15
  • Contact: XIA Daojia E-mail:lee4@nuaa.edu.cn;xdjym@nuaa.edu.cn

摘要:

【目的】 分析人工智能(Artificial Intelligence, AI)作为辅助工具在学术同行评议中的创新应用,提出未来发展建议。【方法】 首先通过文献调研法和案例分析法,比较国内外AI辅助学术同行评议的应用实践;其次按功能对其进行分类,并阐述其主要支撑算法;最后展望AI在学术同行评议领域的未来发展方向。【结果】 国内方面,AI推荐审稿人功能被应用于基金评审,但AI学术影响力预测功能尚停留在研发阶段;国外方面,除学术不端检测功能外,其他功能的应用均处于起步阶段。按功能不同,AI辅助同行评议可分为投稿审查、审稿人推荐和学术影响力预测3类。其未来发展方向为:评议流程和文本结构的标准化、智能算法的通用化、评议专家库和已发表学术论文数据库的集成化;最终AI将发挥评议主体功能。【结论】 科研机构与学术期刊应积极参与全文文献数据库与全球审稿人数据库等相关数字基础设施建设,加快AI在学术同行评议中的应用,进一步推进该领域的数字化和智能化。

关键词: 同行评议, 人工智能, 学术不端检测, 审稿人推荐, 学术影响力预测

Abstract:

[Purposes] This paper aims to summarize the innovative applications of artificial intelligence (AI) in academic peer review and point out the future development directions of AI in academic peer review. [Methods] Through literature research and case study, we compared the practices in AI-assisted academic peer review in China and abroad, and then classified them into three types by function to further describe the main supporting algorithms for each type. On this basis, we estimated the future development directions of AI in academic peer review. [Findings] In China, AI-based reviewer recommendation has been applied to selecting grant reviewers, but the function of AI in academic impact prediction is still being developed. In foreign countries, except for academic misconduct detection, other functions of AI are still in their infancy in application. AI-assisted peer review is classified into preliminary review of papers, recommendation of reviewer, and prediction of academic impact. Standardized review process and paper structure, generalized intelligent algorithms, and integrated reviewer expert database and published academic paper database should be the future development directions of AI-assisted peer review, so that AI can play a key role in peer review. [Conclusions] Scientific journals should actively participate in building digital infrastructures such as full-text database and global reviewer database, thereby speeding up the application of AI in scientific peer review and pushing forward the digitalization and intelligentization of academic publishing.

Key words: Peer review, Artificial intelligence (AI), Detection of academic misconduct, Recommendation of reviewer, Prediction of academic impact