AI time saver!

[Richard Zander]

I had a mixed bag of references I was about to tediously reformat to the specifications of a new journal, when...hey! What about that lazy AI? I sent a sample of the format to "Sparky" who identified it as Vancouver style. When I asked if it would change my references to that, it said "N problem". So...throw away your Zotarro and other reference managers, you can get an AI to reformat all your references in a jiffy. (The DOIs had to be added on by hand.)

[John Hutchinson]

CrossRef provides this tool to add DOIs to a list of references: https://doi.crossref.org/simpleTextQuery .

        John Hutchinson

On Tue, 1 Sept 2026 at 16:39, Richard Zander <finuk...@gmail.com> wrote:

I had a mixed bag of references I was about to tediously reformat to the specifications of a new journal, when...hey! What about that lazy AI? I sent a sample of the format to "Sparky" who identified it as Vancouver style. When I asked if it would change my references to that, it said "N problem". So...throw away your Zotarro and other reference managers, you can get an AI to reformat all your references in a jiffy. (The DOIs had to be added on by hand.)

--
You received this message because you are subscribed to the Google Groups "Taxacom" group.
To unsubscribe from this group and stop receiving emails from it, send an email to taxacom+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/taxacom/161f399f-3cc2-46df-b0a2-d17822416d1bn%40googlegroups.com.

[Richard Littauer]

I would be very careful with using AI to touch references. I've already been on a paper where coauthors introduced hallucinated references, which was disastrous. Formatting can also cause hallucinations. It's not a bad idea to use AI to help with the drudge work of formatting - but a diligent author would need to compare both lists with a red pen after any AI input. 

R

To view this discussion visit https://groups.google.com/d/msgid/taxacom/CA%2B4mfygsLdMigsg%3DwhAusendE0aozvF3ThspzSXfEs%3Dy9kQKwA%40mail.gmail.com.

--

Richard Littauer | burntfen.com | socials: richard.social

[Tyler Wenzel]

As someone who actively looks for ways to be more efficient with AI, I wanted to add a caveat to what Richard said: even reference formatting can be (surprisingly) tricky for LLMs. A few days ago I authored an article where the citation I provided (with a link!) was right, but the AI regenerated the URL I provided it with an API-only access link for the dataset in question rather than a publicly accessible link I provided it. It definitely can help, but it will (at the present) always need a manual check of everything. 

To view this discussion visit https://groups.google.com/d/msgid/taxacom/CACBiX2isgHxrWXzhfJ52sqcMge_S_fH1r7k%3Dpc0VNJ7ZdHJ_-w%40mail.gmail.com.

[Ziv Lieberman]

@Richard Zander, Vancouver is one of the basic styles that comes pre-loaded with Zotero. Why not just use the ref. manager once you've identified the style, which will also reformat everything right away, and can be updated or customized with full knowledge of what changes you're making?

To view this discussion visit https://groups.google.com/d/msgid/taxacom/CAOsLsTtUC9%2BpRiLy3KKmgVLS6_Y3oTkpiM5M39r1a-URNh-NeQ%40mail.gmail.com.

[Glenn Shea]

I've learnt not to trust CrossRef without carefully checking the doi's it sends. For books in particular, it often spits out doi's for book reviews of that book, not the book itself. 

[Tony Rees]

I actually use CrossRef in the other direction - start with a DOI, spit out a formatted reference (this function is invoked behind-the-scenes in the author data entry interfaces of WoRMS and IRMNG, to cut down on manual data entry and improve consistency). Unfortunately as others have noted, the text fields held in  CrossRef against any particular DOI may sometimes be incomplete (especially missing pagination) or just a bit weird (formatting, inconsistent capitalization) in a small but tedious percentage of the time (maybe 1%, but that is still too much to take on board without additional manual checking). 

E.g. this example from a few minutes back as part of my ongoing data entry activities:

10.11646/phytotaxa.736.2.2

authorship comes back as Zhang, Y.-P.; YE, J.; LI, Y.; Huang, M.-R.; Yang, N.; HE, S.-Y.; HE, S.-H.

a little further tweaking required!

Just my experience,

Regards - Tony

[Tony Rees]

Further to my reported experience as above, from a bit more recent intensive experience with CrossRef as a DOI resolver, common problems include incorrect capitalization of article titles (every word capitalized instead of sentence case), missing spaces before or after italicized words, extra line breaks/taps that should not be there in article titles, some incorrect capitalization of author names (unusual), inconsistent treatment of prefixes e.g. Silvs / da Silva etc., missing pagination, occasionally wrong authors altogether or truncated lists (the latter quite rare though), more. Enough to be troublesome... also at frequency greater than 1 in 100 as previously stated - maybe more like 1 in 10 or 1 in 20. So useful, but definitely requiring additional manual review.

- Tony

To view this discussion visit https://groups.google.com/d/msgid/taxacom/45f9d356-4ea0-4a4f-81d3-7855639bb96dn%40googlegroups.com.

[Tony Rees]

An additional inconvenience at times in returned data from CrossRef: dropping second or further initials of authors in lists when they are there in the original work (e.g. P. Crous for P.W. Crous, etc.)...

This is why humans exist, for such checks, of course; or maybe A.I. could do the work for us!!