【目的】 了解“论文工厂”论文(即PM论文)发表和撤回(撤稿、关注、更正)的时间分布、来源期刊、作者及机构的合作关系、撤回原因,以及PM论文的学科分布等情况,为我国科研诚信治理和学术出版质量控制提供参考方案和技术支持。【方法】 从Retraction Watch Database和Web of Science核心合集获取PM论文的撤回数据及其原始文献信息;通过VOSviewer、CiteSpace、Python、Origin等软件对PM论文特征进行文献计量分析和可视化处理。【结果】 截至2021年7月31日,含有“论文工厂”原因的撤回论文共526篇,署名均含中国作者,发表在20家国际出版商的54种期刊上;PM论文的发表和撤回时间分别集中在2017—2019年(占80.42%)和2020—2021年(占98.67%);第一作者来自我国29个省/市的171家单位,出现次数最多的省份和单位分别是山东省(155篇,占29.47%)和吉林大学(57篇,占10.84%);除“论文工厂”外,第三方调查、图片重复使用、结论不可靠3种撤稿原因的标记频次最高;PM论文的学科类别集中在生命科学和健康科学。【结论】 PM论文普遍存在图(表)或数据重复使用、操纵、剽窃或篡改/伪造等现象,论文工厂有组织、大规模的按需造假行为对科学社区构成巨大冲击,严重扰乱出版秩序和科研评价体系,亟需系统地追踪、挖掘、追责和清理。
[Purposes] This paper aims to elucidate the publication and withdrawal (retraction, expression of concern, and correction) time of papers from “paper mill” (PM papers), the source journals, relationship between the authors and institutions, reasons for withdrawal, and discipline distribution of these papers, which is expected to provide reference schemes and technical support for the handling of academic misconducts and academic publishing quality control in China. [Methods] First, the withdrawal data and original paper information of PM papers were retrieved from Retraction Watch Database and Web of Science core collection. Then, VOSviewer, CiteSpace, Python, Origin, and other softwares were employed for bibliometric analysis and visualization of PM paper features. [Findings] As of July 31, 2021, a total of 526 PM papers, on 54 journals of 20 international publishers, had been withdrawn, and the authors of each of the papers all included Chinese scholars. PM papers were mainly published and withdrawn in 2017-2019 (80.42%) and 2020-2021 (98.67%), respectively. The first authors were from 171 institutions in 29 provinces/cities in China and authors from Shandong Province (155 papers, 29.47%) and Jilin University (57 papers, accounting for 10.84%) separately topped the list. Except for PM, the other top three reasons for withdrawal were investigation by third party, duplication of image, and unreliable results. PM papers were mainly about life science and health science. [Conclusions] PM papers feature duplication, manipulation, plagiarism, or falsification/fabrication of images (tables) or data. The organized and large-scale on-demand frauds of PM pose a great impact on the scientific community and seriously disrupt the publishing order and scientific research evaluation system, which should be tracked, mined, handled, and cleaned up in a timely manner.