PLOS Open Science Indicators
收藏DataCite Commons2026-01-05 更新2024-07-29 收录
下载链接:
https://plos.figshare.com/articles/dataset/PLOS_Open_Science_Indicators/21687686
下载链接
链接失效反馈官方服务:
资源简介:
<br>This dataset contains article metadata and information about Open Science Indicators for approximately 139,000 research articles published in PLOS journals from 1 January 2018 to 30 March 2025 and a set of approximately 28,000 comparator articles published in non-PLOS journals. This is the tenth release of this dataset, which will be updated with new versions on an annual basis.This version of the Open Science Indicators dataset shares the indicators seen in the previous versions as well as fully operationalised protocols and study registration indicators, which were previously only shared in preliminary forms. The v10 dataset focuses on detection of five Open Science practices by analysing the XML of published research articles:Sharing of research data, in particular data shared in data repositoriesSharing of codePosting of preprintsSharing of protocolsSharing of study registrationsThe dataset provides data and code generation and sharing rates, the location of shared data and code (whether in Supporting Information or in an online repository). It also provides preprint, protocol and study registration sharing rates as well as details of the shared output, such as publication date, URL/DOI/Registration Identifier and platform used. Additional data fields are also provided for each article analysed. This release has been run using an updated preprint detection method (see OSI-Methods-Statement_v10_Jul25.pdf for details). Further information on the methods used to collect and analyse the data can be found in Documentation.Further information on the principles and requirements for developing Open Science Indicators is available in https://doi.org/10.6084/m9.figshare.21640889.<br><b>Data folders/files</b>Data Files folderThis folder contains the main OSI dataset files PLOS-Dataset_v10_Jul25.csv and Comparator-Dataset_v10_Jul25.csv, which containdescriptive metadata, e.g. article title, publication data, author countries, is taken from the article .xml filesadditional information around the Open Science Indicators derived algorithmicallyand the OSI-Summary-statistics_v10_Jul25.xlsx file contains the summary data for both PLOS-Dataset_v10_Jul25.csv and Comparator-Dataset_v10_Jul25.csv.Documentation folderThis file contains documentation related to the main data files. The file OSI-Methods-Statement_v10_Jul25.pdf describes the methods underlying the data collection and analysis. OSI-Column-Descriptions_v10_Jul25.pdf describes the fields used in PLOS-Dataset_v10_Jul25.csv and Comparator-Dataset_v10_Jul25.csv. OSI-Repository-List_v1_Dec22.xlsx lists the repositories and their characteristics used to identify specific repositories in the PLOS-Dataset_v10_Jul25.csv and Comparator-Dataset_v10_Jul25.csv repository fields.The folder also contains documentation originally shared alongside the preliminary versions of the protocols and study registration indicators in order to give fuller details of their detection methods.<b>Note on the accuracy of the protocols indicator</b> (January 2026)<br>It has come to our attention that the accuracy rates for the protocols indicator as reported in the Protocols_Methods_Statement_Sep23.pdf file are incorrect. The groundtruth exercise for protocols incorrectly identified 19 articles as having protocols in the PLOS corpus and 21 in the Comparator corpus during the manual exercise. These have been corrected in the groundtruth dataset. The accuracy rates for the PLOS and Comparator corpus are therefore 94% and 97% respectively. Documentation will be updated in the next scheduled release (anticipated in July 2026).<br><b>Contact details for further information:</b>Iain Hrynaszkiewicz, Director, Open Research Solutions, PLOS, ihrynaszkiewicz@plos.org / plos@plos.orgLauren Cadwallader, Open Research Manager, PLOS, lcadwallader@plos.org / plos@plos.org<br><b>Acknowledgements:</b>Thanks to Allegra Pearce, Tim Vines, Asura Enkhbayar, Scott Kerr and parth sarin of DataSeer for contributing to data acquisition and supporting information.
<br>本数据集包含2018年1月1日至2025年3月30日期间发表于PLOS期刊的约13.9万篇研究论文的元数据,以及开放科学指标(Open Science Indicators)相关信息;同时纳入约2.8万篇发表于非PLOS期刊的对照论文。本数据集为第十次发布版本,后续将按年度更新迭代。本版开放科学指标数据集延续了此前版本的指标体系,同时新增了正式可操作化的研究方案(protocols)与研究注册指标——此前这两类指标仅以初步形式公开。第十版数据集通过分析已发表研究论文的可扩展标记语言(XML)文件,聚焦于5类开放科学实践的识别:研究数据共享(尤其是通过数据仓储(repository)共享的数据)、代码共享、预印本(preprint)发布、研究方案共享、研究注册信息共享。本数据集提供了数据与代码的生成及共享率、共享数据与代码的存储位置(是否附于补充材料或在线仓储);同时提供了预印本、研究方案及研究注册信息的共享率,以及共享成果的详细信息,如发表日期、统一资源定位符(URL)/数字对象标识符(DOI)/注册标识符及所用平台。此外,还为每篇被分析的论文提供了额外的数据字段。本版发布采用了更新后的预印本识别方法(详细信息参见OSI-Methods-Statement_v10_Jul25.pdf)。有关数据收集与分析方法的更多信息可查阅配套文档。有关开放科学指标开发的原则与要求,可访问https://doi.org/10.6084/m9.figshare.21640889获取。<br><b>数据文件夹与文件</b>
数据文件文件夹
该文件夹包含本数据集的核心文件:PLOS-Dataset_v10_Jul25.csv与Comparator-Dataset_v10_Jul25.csv,二者收录了从论文可扩展标记语言(XML)文件中提取的描述性元数据(如论文标题、发表信息、作者所属国家),以及通过算法推导得到的开放科学指标相关信息;OSI-Summary-statistics_v10_Jul25.xlsx则包含上述两个CSV文件的汇总统计数据。
文档文件夹
该文件夹包含与核心数据文件相关的配套文档:OSI-Methods-Statement_v10_Jul25.pdf阐述了数据收集与分析的底层方法;OSI-Column-Descriptions_v10_Jul25.pdf说明了PLOS-Dataset_v10_Jul25.csv与Comparator-Dataset_v10_Jul25.csv中各字段的含义;OSI-Repository-List_v1_Dec22.xlsx列出了用于识别PLOS-Dataset_v10_Jul25.csv与Comparator-Dataset_v10_Jul25.csv中仓储字段的各类仓储及其特征。此外,本文件夹还包含此前与研究方案及研究注册指标初步版本同步公开的文档,以完整呈现两类指标的识别方法。
<b>关于研究方案指标准确性的说明(2026年1月)</b>
我们注意到,Protocols_Methods_Statement_Sep23.pdf文件中所公布的研究方案指标准确率存在错误。在手动标注的基准真值测试(groundtruth exercise)中,PLOS数据集内被错误判定为包含研究方案的论文共19篇,对照数据集内则为21篇,现已在基准数据集内完成修正。因此,PLOS与对照数据集的准确率分别修正为94%与97%。相关文档将在下一次计划发布(预计2026年7月)中更新。<br><b>获取更多信息的联系方式:</b>
Iain Hrynaszkiewicz,开放研究解决方案总监,PLOS,ihrynaszkiewicz@plos.org / plos@plos.org
Lauren Cadwallader,开放研究经理,PLOS,lcadwallader@plos.org / plos@plos.org<br><b>致谢:</b>
感谢DataSeer的Allegra Pearce、Tim Vines、Asura Enkhbayar、Scott Kerr与Parth Sarin为数据采集及信息支持工作提供的贡献。
创建时间:
2022-12-07
搜集汇总
数据集介绍

背景与挑战
背景概述
PLOS开放科学指标数据集包含约16.7万篇研究文章(2018-2025年)的元数据,专门用于检测和量化五种开放科学实践,包括数据共享、代码共享和预印本发布等。该数据集为第十版,提供完整的指标操作化方法和定期更新,旨在支持开放科学政策的评估与研究。
以上内容由遇见数据集搜集并总结生成



