Login
You're viewing the mastodonapp.uk public feed.
  • Leonardo Fontenelleleonardof@bertha.social
    Oct 15, 2026, 7:59 PM

    Substantive, structured abstracts are here to stay. But people always say abstracts are often the only thing people will (be able to) read from the paper, and I'm not sure this applies anymore.

    It's not like we are looking up physical copies of NLM's Index Medicus anymore. What's more, Latin America has gone open access since online journals became a thing, and the rest of the world has been catching up, with more articles being open access than not.

    #ScholComm

    💬 1🔄 1⭐ 0

Replies

  • Marin T. Kaelmarintkael@mastodon.social
    Oct 16, 2026, 12:50 AM

    There's one reader for whom the abstract is still often the whole paper: the machines. Search crawlers and AI answer tools frequently only get the landing page with title, abstract and metadata, because the full text sits in a PDF they don't parse well or behind a link they never follow. So the abstract quietly turns into the part that gets quoted back to people. Structured abstracts help there too.

    💬 1🔄 0⭐ 0
  • Leonardo Fontenelleleonardof@bertha.social
    Oct 16, 2026, 9:51 AM

    @marintkael I don't use generative LLMs so I don't know about quoting, but Google Scholar indexes the full text and often can find green access PDFs. Ironically MEDLINE relies on abstracts, though, because it decided recently to generate keywords feeding abstracts to a predictive model instead of feeding full text papers to actual people

    💬 2🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • Marin T. Kaelmarintkael@mastodon.social
    Oct 16, 2026, 12:54 PM

    @leonardof @nyhan Fair point on Scholar, it's the one that actually goes and fetches the PDF. The MEDLINE switch is the pattern I meant though: since 2022 NLM's MeSH indexing runs automated on title and abstract. And thanks for I4OA, that's exactly the layer: abstracts deposited openly in Crossref metadata, which is what a lot of tools read when they never open the PDF.

    💬 0🔄 0⭐ 0