RefChecker & Hallucinated Citation Lists

[Michele Gibney]

Do you get contacted by faculty asking to double check citations for hallucination in thesis/dissertations? Is part of your T/D workflow to review the Reference list for accuracy? At my institution, in the library we're only in charge of checking the T/D for formatting or style. We don't check content. However, I have been getting emails lately from faculty on advising committees about hallucinated citations wanting advice on a) how to check for that and b) how to bring it up to the primary advisor and the student. 

I was investigating a tool I saw briefly demo'd by Mark Russinovich in a webinar earlier this year called RefChecker - https://github.com/markrussinovich/refchecker

I have set it up on my computer for testing purposes but it's currently using my API keys (and therefore my credit card) for the LLM use portions of the checks - so this is not a sustainable model and I would need our IT department to set up something for longer term use and have buy in from all graduate programs. And it does a semi-decent job although the last Reference list I fed it came back with 2 hallucinated citations both of which are dissertations in my institutional repository (not hallucinations) - so it still needs human oversight.

Basically, I'm just wondering if this is a problem others are encountering in your oversight roles of T/Ds and what you might be doing to address it. Has anyone else heard of/is using RefChecker? Are you using anything else for this purpose? - and I'm meaning specifically Reference lists, not the full manuscript. We're not looking for an AI writing detector just a citation hallucination checker.

Thanks!

Best,

Michele

Michele Gibney, PhD, MLIS

Head of Publishing and Scholarship Support
University Libraries

UNIVERSITY OF THE PACIFIC
3601 Pacific Ave. | Stockton Campus Library | Stockton, CA 95211  

209.946.2785 | mgi...@pacific.edu

pacific.edu/library

[Institutional Repository Coordinator]

I would think this would be the responsibility of the advising committee - not the library.   

[Bailey, Jody Elizabeth]

Hi Michele,

 

My office manages Emory’s ETD repository, and we do not check ETD submissions for formatting, style, or hallucinations in the reference list. All content (I assume that includes any problems with the reference list) is reviewed and approved (or not) by the student’s committee, and formatting is checked by approvers in their respective schools. I would not want to get into the business of having to review anything to do with content. IMO, that is the purview of the subject experts since they are familiar with the literature in their discipline.

 

Best regards,

 

Jody

 

Jody Bailey, MA, MLIS (she, her)

Head, Scholarly Communications Office

Emory University

jody....@emory.edu

Schedule an appointment with me

https://orcid.org/0000-0002-4226-4173

I sometimes work flexibly and send emails outside normal office hours.

No need to respond to my emails outside yours.

 

--
You received this message because you are subscribed to the Google Groups "ETD" group.
To unsubscribe from this group and stop receiving emails from it, send an email to etd+uns...@ndltd.org.
To view this discussion visit https://groups.google.com/a/ndltd.org/d/msgid/etd/27247db9-79a9-43e2-949d-a599f7cf23e0n%40ndltd.org.

[Jason Skoog]

Hi Michele,

I don't review the papers for content. When the papers arrive to our repository, they are considered approved by the faculty, and I accept what the faculty provides me. We have the following withdrawal policy in place:

"Works may be withdrawn from the Institutional Repository for reasons including but not limited to copyright violations, other legal issues, plagiarism, factual inaccuracies, and author request."

However, I don't actively check for these. If someone brought one of these issues to my attention, I would alert the program, who would investigate.

On Wednesday, March 25, 2026 at 10:47:46 PM UTC-5 Michele Gibney wrote: