Improving the Scopus and Aurora queries to identify research that supports the United Nations Sustainable Development Goals (SDGs) 2021

The United Nations Sustainable Development Goals (SDGs) challenge the global community to build a world where no one is left behind. Since 2018, Elsevier have generated SDG search queries to help researchers and institutions track and demonstrate progress towards the targets of the United Nations Sustainable Development Goals (SDGs). At the end of 2018, Elsevier worked on 2 versions of the SDG queries. One version was created by the Elsevier Analytical Services group and another by the Science-Metrix group, who had recently become part of Elsevier. At that time Science-Metrix was creating queries for 5 of the 16 SDGs, as part of pro-bono work for UNESCO. In 2020 inspired by the earlier queries, Elsevier, through its Science-Metrix group, used a new approach to mapping publications to the SDGs. Taking customer feedback into account, they significantly increased the number of search terms used to define each SDG. Those queries were then complemented by a machine learning model, which helped increase the recall by approximately 10%. As a result, this year’s “Elsevier 2021 SDG mapping” captures on average twice as many articles as the 2020 version, while keeping precision above 80%. The mapping also has a better overlap with SDG queries from other independent projects. Times Higher Education (THE) are using the “Elsevier 2021 SDG mapping” as part of their 2021 Impact Rankings. The documentation below describes the methods used and shares the queries. For each SDG, you can download the query as a text file, along with an html file that describes the methodology used to create the search query, plus additional information such as the most influential keyphrases and journals. It also breaks down the query into digestable chunks. A separate folder contains the methodology for the machine learning component, along with a sample of the top 100 keyphrases per SDG and a stratified sample of 8,000 EIDs that the model identified arcoss the SDGs.

联合国可持续发展目标（Sustainable Development Goals, SDGs）号召全球社会携手构建一个无人被落下的美好世界。自2018年起，爱思唯尔（Elsevier）推出SDG检索式，助力研究者与科研机构追踪并展现其在落实联合国可持续发展目标各项指标上的进展。
2018年末，爱思唯尔共推出两版SDG检索式：一版由爱思唯尔分析服务团队打造，另一版则由彼时刚并入爱思唯尔的科学计量（Science-Metrix）团队开发。当时Science-Metrix正为联合国教科文组织（UNESCO）开展公益项目，仅为16项SDGs中的5项构建检索式。
2020年，受此前检索式启发，爱思唯尔通过其Science-Metrix团队采用全新方法将学术文献映射至SDGs。结合用户反馈，他们大幅扩充了每项SDG对应的检索词数量。随后辅以机器学习模型，使查全率提升约10%。
最终，2021年版“爱思唯尔SDG映射工具”平均收录文献量为2020年版的两倍，同时精准率维持在80%以上，且与其他独立项目的SDG检索式重合度更优。泰晤士高等教育（Times Higher Education, THE）已将“爱思唯尔2021 SDG映射工具”纳入其2021年影响力排名评估体系。
下文将详述所用方法并共享相关检索式。针对每项SDG，用户可下载检索式文本文件与阐释检索式构建方法的HTML文档，其中还包含最具影响力的关键词短语与期刊等补充信息。该文档还将检索式拆解为便于理解的模块。
另有独立文件夹收录机器学习模块的方法论文档，以及每项SDG排名前100的关键短语样本，还有模型在全SDGs范围内识别出的8000条经过分层抽样的EID样本。