fighting paper mills
Glenn Hampson
From LinkedIn, here’s more interesting info from James Butcher today: https://bit.ly/48CC6Se. And here’s the full text in case you aren’t on LinkedIn:
Scholarly publishers' content will come under increasing scrutiny in 2024.
The open access gold rush is over. There are new sheriffs in town.
Last week's APE meeting demonstrated the power of tools like Signals and Clear-Skies (developed by Elliott
Lumb and Adam
Day) to identify papers that may have research integrity issues.
Crucially, the tools can also be used to assess the trustworthiness of content published at the journal and portfolio level.
A recent preprint by Simon Porter and Leslie
McIntosh from Digital
Science provides another way to identify articles that come from paper mills (see: https://lnkd.in/ezhptSCn).
As generative AI improves, fake text will become harder to spot. Detecting unusual clusters of authors is likely to be a more viable approach.
For example, a paper that has 5 authors, none of whom has published before, is a red flag. Author networks are harder for paper mills to fabricate than text.
Figure 6 in the preprint shows the proportion of content from 10 leading publishers that have likely been generated by paper mills.
IOP Publishing is
the only publisher that is getting better, not worse, at spotting the fakes over time.
Publishers risk reputational damage if they publish fraudulent articles. If submissions fall, so will revenues, especially on open access journals.
Investment in research integrity tools and teams is mission critical.
What other tools should be on publishers' radar? Please leave a comment below.
Kaveh Bazargan
- I know both Adam and Elliott well and they are doing good work. Elliott's work, Signals, is the most interesting for me. You can try it yourself by entering a DOI.
- There is also Guillaume Cabanac's Problematic Paper Screener that finds so-called Tortured Phrases. These are deliberately reworded text to get around plagiarism checkers, leading to ridiculous phrases such as "fluffy rationale" instead of "fuzzy logic"!! If you think it's not a problem look at 970 papers already published with this phrase.
- Then there is the huge area of figure manipulation. This is where we have been most active, supporting the great sleuths such as Elisabeth Bik. We have produced some 1600 animations in this area. In case you are interested here is a recent presentation from me specifically on image misconduct. (I will also be presenting at the R2R conference in Feb.)
- Generative AI is going to compound the problem for sure, both in text and in images. Here is an example I gave regarding PhotoShop Generative AI.
- IOP Publishing have been very good at dealing with the problem, and have even announced they will be donating the original APCs for retracted articles to charity!
Kaveh
--
As a public and publicly-funded effort, the conversations on this list can be viewed by the public and are archived. To read this group's complete listserv policy (including disclaimer and reuse information), please visit http://osinitiative.org/osi-listservs.
---
You received this message because you are subscribed to the Google Groups "The Open Scholarship Initiative" group.
To unsubscribe from this group and stop receiving emails from it, send an email to osi2016-25+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/osi2016-25/DM4PR17MB6064DD2CC288EB7C6C06AEB9C5732%40DM4PR17MB6064.namprd17.prod.outlook.com.
Rob Johnson
Rob Johnson (he/him)
Managing Director
Follow us on Twitter @rschconsulting
M: +44(0)779 511 7737
E: rob.j...@research-consulting.com
W: www.research-consulting.com