Musings on MeSH

Posted by Erica Lake on December 22nd, 2023 Posted in: Blog

---

Around here, January is MeSH Month.

It’s when NLM and NNLM staff highlight the new and updated MeSH terminology, offer training on how your PubMed searches may be affected by the MeSH updates, and answer your MeSH and indexing questions. Mark your calendar for our 2024 sessions – if you use PubMed, you won’t want to miss them!

 

NLM Webinar: 2024 MeSH Highlights
January 10, 12-1pm CT
Register

NNLM Class: MeSH Changes and PubMed Searching
January 25, 12-1:30pm CT
Register

 

In preparation for these sessions, I found myself reflecting on the MeSH conversations I’ve had this year. Several librarians expressed their disappointment with automated indexing, and some shared specific examples of inaccurately assigned MeSH terms. The plenary session at the MLA Midwest Chapter Conference in October further delved into this concern.

Speaker Alexandre Amar-Zifkin presented findings from a study he and his colleagues at the Université de Montréal conducted. While teaching PubMed searching to students, they had experienced embarrassing moments of finding inaccurate or missing MeSH for key concepts of an article. They decided to review a sample of records with automated indexing from early 2023 to determine whether their main concepts were adequately represented with MeSH. They developed a methodology, ran the review, and concluded that 47% of the records had minor or major MeSH issues that would have affected their likelihood of retrieval. The team shared their findings and their data with NLM, and corrections have been made to the problematic citations.

NLM is continually developing the algorithm that is used for indexing, and this study as well as other internal studies, contribute to that process. The current algorithm is different than what was in place when Amar-Zifkin‘s study was conducted. What’s more, an updated algorithm is planned to be released this January with a focus on precision improvements. Future updates will focus on recall and on correctly capturing MeSH Supplementary Concept Data. Other future work will investigate automated indexing based on processing of the article’s full text (where NLM has access to that text for computational purposes). Currently, automated indexing is based on the title and abstract. You can learn more on NLM’s Automated Indexing FAQ page.

Can we still count on MeSH? I believe so.

MeSH remains a crucial tool. Relying solely on human indexing is no longer practical, and the continuous refinement of the algorithm underscores the commitment to accuracy.

When teaching, the emphasis remains on leveraging both MeSH and text words to retrieve comprehensive results. Even before automated indexing, NLM advised not to rely exclusively on MeSH for comprehensive searches because you would miss the unindexed.

Researchers and authors need to ensure their abstracts and titles are clear, and this could be a teaching point for librarians who help authors prepare their manuscripts for publication. Since early 2013, PubMed has displayed publisher-supplied keywords, providing authors with an avenue to incorporate preferred MeSH terms for better discoverability. Authors can identify MeSH terms by consulting the MeSH Browser or using tools like MeSH on Demand. If authors wish their articles to be retrieved by their preferred terminology, those words need to appear in the title or abstract.

NLM and NNLM still count on you, and actively encourage your feedback. Users are urged to report problematic indexing through the “Help” link available at the bottom of every NLM webpage. This will take you to the NLM Support Center where you can contact the NLM Helpdesk with your feedback and questions. There is also a MeSH suggestion page with specific directions on how to submit MeSH suggestions, updates and corrections. These direct lines of communication allow librarians, researchers, and authors to contribute to the ongoing improvement of MeSH indexing.

Happy MeSH Month!