A journal has allowed a geophysicist who cited his own work hundreds of times across 10 papers to retract the articles and republish them with a fraction of the self-citations.
From 2017 to 2019, Yangkang Chen published some of the papers in Geophysical Journal International, an Oxford University Press title, while he was a postdoc at Oak Ridge National Laboratory in the U.S., and some as a faculty member at Zhejiang University in China. In April, the journal subjected the works to expressions of concern.
On June 19, the journal published a retraction notice for the 10 papers, along with an editor’s note that read:
The Editor-in-Chief was recently alerted to concerns of citation stacking in the following articles:
…
The author subsequently admitted to participating in this practice and apologised for his misconduct in this instance. He has since revised the articles to exclude these extra citations, which did not affect the conclusions presented. The new versions have been reviewed by members of the journal’s Editorial Board and have been republished after retraction of the original version of the manuscript, at the request of the author:
Here are the ten papers, and their replacements, according to the journal:
- ‘Fast dictionary learning for noise attenuation of multidimensional seismic data’ by Yangkang Chen, Geophysical Journal International (2019) 209, 21 (doi.org/10.1093/gji/ggw492) will be replaced by ‘Fast dictionary learning for noise attenuation of multidimensional seismic data’ by Yangkang Chen, Geophysical Journal International (doi.org/10.1093/gji/ggaa184)
- ‘Low-frequency noise attenuation in seismic and microseismic data using mathematical morphological filtering’ by Yangkang Chen, Geophysical Journal International (2017) 211, 1296 (doi.org/10.1093/gji/ggx371) will be replaced by ‘Low-frequency noise attenuation in seismic and microseismic data using mathematical morphological filtering’ by Yangkang Chen, Geophysical Journal International (doi.org/10.1093/gji/ggaa185)
- ‘Automatic microseismic event picking via unsupervised machine learning’ by Yangkang Chen, Geophysical Journal International (2018) 212, 88 (doi.org/10.1093/gji/ggx420) will be replaced by ‘Automatic microseismic event picking via unsupervised machine learning’ by Yangkang Chen, Geophysical Journal International (doi.org/10.1093/gji/ggaa186)
- ‘Seismic noise attenuation using an online subspace tracking algorithm’ by Yangkang Chen, Geophysical Journal International (2018) 212, 1072 (doi.org/10.1093/gji/ggx422) will be replaced by ‘Seismic noise attenuation using an online subspace tracking algorithm’ by Yangkang Chen, Geophysical Journal International (doi.org/10.1093/gji/ggaa187)
- ‘Plane-wave orthogonal polynomial transform for amplitude-preserving noise attenuation’ by Yangkang Chen, Geophysical Journal International (2018) 214, 2207 (doi.org/10.1093/gji/ggy267) will be replaced by ‘Plane-wave orthogonal polynomial transform for amplitude-preserving noise attenuation’ by Yangkang Chen, Geophysical Journal International (doi.org/10.1093/gji/ggaa188)
- ‘Deblending of simultaneous-source data using a structure-oriented space-varying median filter’ by Yangkang Chen, Geophysical Journal International (2019) 216, 1214 (doi.org/10.1093/gji/ggy487) will be replaced by ‘Deblending of simultaneous-source data using a structure-oriented space-varying median filter’ by Yangkang Chen, Geophysical Journal International (doi.org/10.1093/gji/ggaa189)
- ‘Five-dimensional seismic data reconstruction using the optimally damped rank-reduction method’ by Yangkang Chen, Geophysical Journal International (2019) 218, 224 (doi.org/10.1093/gji/ggz130) will be replaced by ‘Five-dimensional seismic data reconstruction using the optimally damped rank-reduction method’ by Yangkang Chen, Geophysical Journal International (doi.org/10.1093/gji/ggaa190)
- ‘Learning the blending spikes using sparse dictionaries’ by Yangkang Chen, Geophysical Journal International (2019) 218, 1379 (doi.org/10.1093/gji/ggz200)will be replaced by ‘Learning the blending spikes using sparse dictionaries’ by Yangkang Chen, Geophysical Journal International (doi.org/10.1093/gji/ggaa191)
- ‘Least-squares decomposition with time-space constraint for denoising microseismic data’ by Yangkang Chen, Geophysical Journal International (2019) 218, 1702 (doi.org/10.1093/gji/ggz145)will be replaced by ‘Least-squares decomposition with time-space constraint for denoising microseismic data’ by Yangkang Chen, Geophysical Journal International (doi.org/10.1093/gji/ggaa192)
- ‘Automatic high-resolution microseismic event detection via supervised machine learning’ by Yangkang Chen, Geophysical Journal International (2019) 218, 2106 (doi.org/10.1093/gji/ggz273) will be replaced by ‘Automatic high-resolution microseismic event detection via supervised machine learning’ by Yangkang Chen, Geophysical Journal International (doi.org/10.1093/gji/ggaa193)
A look at the references in each paper tells the story. In the original version of “Fast dictionary learning for noise attenuation of multidimensional seismic data,” Chen cited 24 of his own papers. In the republished version, that number was 10. For “Low-frequency noise attenuation in seismic and microseismic data using mathematical morphological filtering,” the figures were 47 and then 11. For “Low-frequency noise attenuation in seismic and microseismic data using mathematical morphological filtering,” it was 33 vs. 7.
In all, Chen cited himself 370 times in the ten original papers — and is down to an apparently more acceptable 105 times in the new versions.
Neither Chen nor the editor of the journal responded to our requests for comment.
What’s unclear is whether the move will have any effect on Chen’s Google Scholar profile, or h-index, both of which rely on the number of times his work has been cited. As best we can tell, Google Scholar has no way of discounting citations.
Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on Twitter, like us on Facebook, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].
When you write your article, of course you cite yourself. It not you, who else? Second point, when you try to publish a paper in open access journal, citing non open journals does not seem logical. This may make big bad journals angry.
Note that “In all, Chen cited himself 370 times in the ten original papers”. This is excessive. I don’t see how it can be justified. Chen was clearly doing to game the metrics for personal gain, not for scientific purposes.
I teach graduate students that if >25% (granted, arbitrary but you have to draw a line somewhere) of references in your paper are self-citations, then that shows to the reader that either a) the work does not have a broad scope (i.e. is of limited interest to only its author) or b) the literature survey was poorly conducted, inappropriately omitting references to the foundational work of others.
It is my personal opinion that we should strive for Newton’s ethos of “standing on the shoulders of giants”, rather than “of course you cite yourself. It not you, who else?”
I worked for years for someone who was pretty clearly #1 in his field. He didn’t cite 47 of his own papers in any study he wrote. He wrote a book on the field which is extremely thorough and comprehensive, and *that* doesn’t cite 47 of his papers. I can’t imagine how you’d come to do so. *Maybe* if you were in one of the subfields where everyone’s name is on every paper (I understand that some parts of astronomy and particle physics are like that), but even so, this number is staggering.
(I am presuming here that “47 self-citations” means “47 papers of mine cited” and not “47 mentions of my papers.”)
If you don’t cite yourself, then who will?
I can’t believe this is a serious question from a serious person. Others. The answer you’re looking for is “others”, “your peers”, “your fellow scientists”.
Your work should be compelling enough and accurate enough that others could and will use it.
Science is team game, as in your peers read your work and build on it or not.
One idea to reduce this self-puffery would be for editors to read the papers BEFORE accepting for publication, and maybe even get manuscripts reviewed by uninvolved peers. It could be called “peer review” or something like that. Wonder if they’ve considered it.
If the citation is relevant, self-citation should not be a problem (even if the number is high). You are citing works that were previously peer reviewed and accepted. I see the problem when it is just filler to increase your citation numbers (which seems the case), as removing those citations does not have any effect on the paper.
If you are the world authority on rare disease X, sure, I could see it. (Though the high numbers here are absurdly high even for that.) Most of us are not in that situation. I worked in a pretty small subfield, but it was still never germane to cite more than 5-6 of my own papers.
The issue is not the number of citations, it’s their relevance. I’ve written papers with 20 self citations because I’ve done most of the large national studies in my area, so any time I report a new national study it builds on the previous ones. So that’s 10-15 citations right there.
If people are worried about it having some impact on career success, then there should be separate stats for total cites and self-cites.
If irrelevant cites were being stuck in just to boost h-index, it was good of the journal to address it. If they were relevant to the topic and he had to cut relevant self citations just to get it published, then the journal was in the wrong.
In my case , i come out with one simple cavity to test the thin film as a pH sensor , i had published many papers using this design in the end i did one paper to share the researcher how simple this design how easy to use , how non distractive method using this cavity, and to show the evidence of this idea i have cited all my papers that used this cavity but this paper got rejected many times because of my self citation
His Google Scholar profile lists 144 papers cited more than 10 times, with a total h-index of 46. He got his PhD in 2015. How does this compare to others in his field? It seems an unusually fast rise but I am not familiar with practices in geophysics.
Simply counting a number is meaningless. It’s more reasonable to evaluate the ratio of the self-citations (10%,30%,50%, and 80% are completely different), the article type (regular article or review article), and the field of the study (wether the field is small and wether the author is an authority in that field).
I’m not quite sure what the problem is here. If these papers were reviewed and assessed as worth publishing initially, and the only problem identified with them post publication is improperly excessive self-citation, why would they not be equally worth (re)publishing after the excess self-citation is removed?
I am a member of the Royal Astronomical Society (RAS), which publishes Geophysical Journal International (GJI), and I looked into this matter a bit. (I have published many papers in GJI and its predecessor GJRAS in the past 46 years.)
It appears that the editors relied on the COPE retraction guidelines.
https://publicationethics.org/retraction-guidelines
I quote below:
“Retractions are not usually appropriate if:
• The main findings of the work are still reliable and correction could sufficiently address errors or concerns”
The issues here are subtle, and I have some sympathy for the editors, although I believe their decision was, ultimately, wrong. Their argument seems to be (1) that the findings reported in the paper were correct, and that correction (removing the egregiously excessive self-citations) could sufficiently address concerns; (2) the COPE guidelines did not allow retraction.
As for (1), I haven’t read the papers (I would if someone in authority asked me, but not otherwise, and I doubt that anyone will ask me), so I won’t offer an opinion on this point, other than skepticism in general. As for (2), the issue hangs on the word “usually” in the COPE guidelines (“Retractions are not USUALLY appropriate if…”). COPE doesn’t say, “”Retractions are NEVER appropriate if…” So the question is whether, in the event of egregious and repeated efforts to game the system by excessive self-citation, editors are justified in enforcing retraction. I say “yes,” but much of the blame for this debacle should be attributed to the fuzziness of COPE’s guidelines for retraction, rather than being solely directed at the editors of GJI.
“No harm, no foul” applies in professional basketball. On the other hand, if repeated excessive self-citation is a no-lose proposition, as it was in this case, what’s to stop it?
I hope COPE will soon put more teeth into their guidelines.
It appears to me that all these papers are closely related. Does anyone find that all these titles are more or less connected with the same key words (noise attenuation, denoising, signal detection) ? If the author is a world authority in this field, does that mean all these retraction are not appropriate? How can’t the reviewers find the citation excessive before recommending the acceptance? Are they qualified? Or do they just think these citations are reasonable? Besides, several other questions remain, like why exactly ten papers are retracted? why other papers from the author are fine? why these papers span three years?
The retractions are weird to me.
I really didn’t get the point here.
Self-citation is really something that seems ”not good” but unavoidable. What really matters is wether the citation is relevant or not, which should not be judged simply by the number but instead should be judged by the peers, e.g., the reviewers. As long as the reviewers think that the citations are reasonable and let it pass, I don’t think there should be any serious disputation on wether it is suitable or already crossing the line.
I saw that many material (or some other domains) scientists have an incredible number (60/80/even higher) of self-citations in one single paper. I even saw many review papers cite the work from someone or the author himself by an even higher number or ratio. The number (or even the ratio) is something tricky to define when writing papers or doing editorial screening. I saw many occasions when an author intentionally avoid citing their own work even if they are crucial or when an editor requests an unreasonable reduction of the authors’ own references even if they are all closely related to the research.
We often heard of a term called ”coerced citation”, which indicates that the reviewers (or editors) enforce the authors to cite an excessive number of papers from them that are not relevant. It is clearly a misconduct because it is an obvious intention to simply boost the citation record of the reviewers and the relevance of the citations are not judged by other independent experts. The self-citation, however, is a completely different thing. Some people complain about the rejection of their papers because of their excessive self-citation. I think it is reasonable to reject a paper during the review process because of that, since obviously reviewers have found these citations not appropriate or relevant. What if the reviewers consider the self-citations as relevant and appropriate? Then what is the criterion to define an excessive self-citation? By the number or by the ratio?
Even if the self-citations are indeed excessive as found by the author himself later after the publication, I didn’t see any inappropriateness for the fact that the author apologize and request a correction. As long as the scientific conclusion is correct, and the correction is only used to remove the unsuitable citations and to make the paper presented in a more clear and more appropriate format, why should not it be considered as a good way?