澳大利亚昆士兰科技大学研究团队利用人工智能工具对260万篇1999至2024年发表的癌症研究论文进行了分析，识别出超过25万篇具有可疑特征、可能源自学术论文工厂的研究1。该AI系统通过检测文本模式中的异常特征，能够以91%的准确率识别出这些可疑论文1。
研究发现，可疑论文的比例呈现上升趋势1。从2000年代初的约1%上升到2022年已超过16%1。昆士兰科技大学Adrian Barnett教授指出："论文工厂是那些出售虚假或低质量科学研究的公司，它们在以工业规模生产'研究'"1。目前已有三家科学期刊开始在编辑审查流程中测试该AI系统1。该研究已发表在《英国医学杂志》(The BMJ)1。
Researchers at Queensland University of Technology in Australia have deployed an artificial intelligence system to identify fraudulent cancer studies on a massive scale. After analyzing 2.6 million cancer research papers published between 1999 and 2024, the team's machine learning tool flagged more than 250,000 papers with suspicious characteristics indicative of origin from academic paper mills 1. The AI system achieved a 91 percent accuracy rate in detecting these problematic papers by identifying abnormal text patterns 1.
The proliferation of fraudulent research in oncology has accelerated sharply in recent years, with the proportion of suspicious papers rising from approximately 1 percent in the early 2000s to over 16 percent by 2022 1. According to Adrian Barnett, a professor involved in the research, "Paper mills are companies that sell fake or low-quality scientific studies. They are producing 'research' on an industrial scale" 1. The detection tool is already undergoing practical testing, with three scientific journals now integrating it into their editorial review workflows 1. The findings were published in The BMJ in 2026 1.
评论
还没有评论，欢迎留下第一条。