The aim of this study is to explore trends in retracted publications in life sciences and biomedical sciences over axes like time, countries, journals and impact factors, and topics. Nearly seven thousand publications, which comprise the entirety of retractions visible through PubMed as of August 2019, were used. This work involved sophisticated data collection and analysis techniques to use data from PubMed, Wikipedia, and WikiData, and study it with respect to the above mentioned axes. Importantly, I employ state-of-the-art analysis and visualization techniques from natural language processing (NLP) to understand the topics in retracted literature. To highlight a few results, the analyses demonstrate an increasing rate of retraction over time and noticeable differences in the publication quality (as measured by journal impact factor) among top countries. Moreover, while molecular biology and cancer dominate retractions, we also see a number of retractions not related to biology. The methods and results of this study can be applied to continuously understand the nature and evolution of retractions in life sciences, thus contributing to the health of this research ecosystem.

这项研究的目的是探索生命科学和生物医学领域中撤回的出版物的发展趋势，涉及时间，国家，期刊和影响因素以及主题等方面。截至2019年8月，已使用近7,000种出版物，其中包括可通过PubMed看到的全部撤回文献。这项工作涉及复杂的数据收集和分析技术，以使用来自PubMed，Wikipedia和WikiData的数据，并针对上述轴进行研究。重要的是，我采用自然语言处理（NLP）的最新分析和可视化技术来理解收回文献中的主题。为了突出一些结果，分析表明，随着时间的推移，撤稿率不断提高，并且排名靠前的国家/地区之间的出版质量（按期刊影响因子衡量）存在明显差异。此外，尽管分子生物学和癌症主导着退缩，但我们也看到了许多与生物学无关的退缩。这项研究的方法和结果可用于不断了解生命科学领域回缩的性质和演变，从而为该研究生态系统的健康做出贡献。