Medical Libraries are all out of date. LLMs still lie and are not regulated
I can sort it out. I was writing more to myself to write down my intentions. MESH has a long pedigree and its interactions with other classifications, complex and filled with many distractions and stories. I will look at the whole history of MESH again and study the various languages together.
Thank you for replying.
Things are a bit scattered. Or there are just many red herrings.
https://mesh.inserm.fr/FrenchMesh/ – https://mesh.inserm.fr/FrenchMesh/conditions_fr.htm
https://mesh.kib.ki.se/
https://decs.bvsalud.org/
==> https://ebooks.iospress.nl/doi/10.3233/SHTI220653, Usability Evaluation of a Modern Multilingual MeSH Browser ==> download link inaccessible by simple methods.WikiMESH at https://pubmed.ncbi.nlm.nih.gov/35612105/
–> Medical Text Indexer
–> DeepMeSH
–>MESHNow
Every country, every medical school, every medical library, every journal, every publisher, and many nonprofits, hospitals and others. Subject headings do not answer questions and they are blunt pointers that mostly say “find something over there to read, read it yourself, and figure it out your self.(“medical AIs”) has 21,400 entry points
(“medical AIs” “MESH”) has 88 entries, mostly not very relevant
It is like a party where every page wears a strange different costume. Each search page is completely different, not just the colors and languages, but it seems the purpose. I cannot tell for sure. Compared to the newer parts of the Internet, these old DOI pages are clumsy and often broken or obscure.I have pretty much given up on Library of Congress and libraries in general with respect to the Internet. They all sit inside their four walls and do not look at the whole world as one network where all libraries work as one – and do not use it to enrich themselves or demand taxes.
Automatic Annotation of PubMed Articles with MeSH Qualifiers (Jul 2023) at https://pubmed.ncbi.nlm.nih.gov/38082894/
(“LLMs” “pubmed”) has 60,100 entry points, but there is no clear human oversight for the training and use of LLMs that are expensive for large use, often falsify answers, cannot trace sources and cannot or will not trace logic. I recommend that NO LLM be trusted where human life or property are involved. Even “medical trials” are problematic, because even very learned and wise humans can be tricked by slick LLM wrapper programs (LWPs) since those groups, none of them are regulated or publicly audited and verifiable.I have a plan now, but I will just use what I already have. Many of the old methods are too costly anyway, when sensors and AIs, and the Internet are changing the way humans treat knowledge. Not as something stored in musty building and dusty books. Or even in vaults and storehouses or committees. The whole of this LLM based AI craze can be corrected, but I expect a few thousand people will die or be mistreated first for it to rise to consciousness of old systems.
Most of these articles are out of date. A few months is a generation now. I have no power. I can only recommend and suggest. I have tried to tell Elon Musk to hurry and do it right, but he hired really young people with no subject matter experience, only computing and a very narrow part of that. The others are going to take decades – they get paid whether they do anything or not.
