25

I am writing my PhD thesis now. It comprises, in particular, one published article in a Q1 journal. This is a compilation thesis, and thus articles are pasted directly. For context, we apply machine learning techniques to biology, and publish in biology journals.

Here is the issue: When trying to solve a problem, I upgraded a machine learning method with a modification. Since, at the time, I was not good at all at reviewing literature, I thought that no one had had this idea before, and I called the method X. Now, while writing the thesis document and properly reviewing literature, I just discovered that this methodology already exists in the machine learning field, and it is called Y (which, by the way, is a better name). In addition, several more advanced adaptations also exist.

What is the best way to proceed here? The thesis has an introduction, and I am thinking maybe citing the article describing Y and saying we do the same thing and call it differently, I think it is the best course of action. I am deeply worried that I did not really contribute with anything new and concerned that someone may think that this is plagiarism.

Of course, none of the co-authors (it is just three of us), nor the reviewers, noticed this.

4
  • Yes, this is a PhD thesis. Corrected, thanks.
    – user17238
    May 13, 2023 at 11:15
  • 5
  • 3
    The, to me, most striking example of not-actual-novelty is the Fast Fourier Transform, which is commonly attributed to Cooley and Tukey (1965) but was in fact already used all the way back in 1805 by Gauss! May 15, 2023 at 10:50
  • 3
    @leftaroundabout ... and finally named after J Fourier :)
    – WoJ
    May 15, 2023 at 16:33

4 Answers 4

50

I would acknowledge that you've since found that someone else did something similar/the same thing. You can still describe how you originally came to your findings and the checks/tests you performed independently.

Independent discovery is not plagiarism when you don't know about the previous finding. Hiding that someone else already found something you thought was original is quite suspicious though. Now that you know, act as though you know.

It's common that things are discovered more than once. Communication barriers used to play a big role, through language, physical distance, and distance in time (things get forgotten or their applications are not evident immediately). Today, many of these barriers are reduced, but instead there is simply so much information out there that it's easy for things to be lost in the background, especially when terminology differs.

6
  • 12
    describe how you originally came to your findings and the checks/tests you performed independently. ... #rediscovery. #reproducibility May 13, 2023 at 7:46
  • 2
    +1 for that last paragraph. One famous example from communications: BCH codes were named after the inventors Raj Chandra Bose, D.K. Ray-Chaudhuri and Alexis Hocquenghem, however, the latter (Hocquenghem) discovered them indepedently of the other two.
    – Sabine
    May 14, 2023 at 9:30
  • 1
    +1 for "especially when terminology differs" - more than once, I have come across papers that described, on an abstract level, the same thing I was about to write, but that used a totally different vocabulary than the literature I was basing my work on. May 14, 2023 at 14:50
  • 1
    is another important point to mention that it may be up to their PhD committee to say if the work still qualifies towards the degree ?
    – Mike M
    May 14, 2023 at 15:05
  • 2
    @MikeM It really should, but certainly not all PhD committees/graduate programs are reasonable, I've seen some really unreasonable policies mentioned on this site.
    – Bryan Krause
    May 15, 2023 at 15:49
27

Change your claim. Instead of 'This thesis proposes a novel method X for ...' to 'This thesis presents the first study that shows how method Y can be applied to field Z'.

The strategy is to present a new angle (or research question) or details. Usually, this is 'easy' (though requires a bit of thought) because you and the author of method Y took a different path/direction or research aim.

3
  • 7
    +1 ... shows how method Y can be applied to field Z May 13, 2023 at 7:45
  • 4
    Since method X was already published, I think it is fair to state in the introduction of the thesis that this was an independent development. Of course one should state in the discussion that it is essentially method Y, but also that this connection was only realized later.
    – silvado
    May 15, 2023 at 12:02
  • 2
    Honesty is always a win. While this means to "downgrade" one's claim from "we developed the novel method X" to "we apply method Y to field Z"; one can, especially in a thesis which is not page/word limited, discuss the whole process: we developed method X and later realized that X is essentially equivalent to the method Y; and here's why this is the case, and here's why X/Y is better than Y/X, etc. etc. No matter whether your method X is novel or not; you developed it, tell the world about it.
    – Dohn Joe
    May 15, 2023 at 16:20
16

This type of thing is not as unusual as you might think; rediscoveries are common and sometimes they slip into the literature because the relevant parties are not aware of the previous work. Just be clear and honest in your dissertation about what happened, and note it where relevant when citing this work in future.

Going forward, you should refer to the relevant method as Method Y (since it was first and you say that this is the better name) and cite both papers, but perhaps add a footnote noting that your own paper was an independent rediscovery of the method and noting why you've chosen to go with the previous method name. This is not plagiarism, but you are correct that this circumstance reduces the novelty of your own work. Your paper might still have value in presenting the same method in a different way and that would still be a novel contribution in a different sense. In any case, keep pushing forward with your research and pivot to new forms of novel contribution in your field.

2
  • 14
    If nothing else, it provides value from confirming the results of the other work -- something entirely too rare in science.
    – Mark
    May 13, 2023 at 6:37
  • 3
    @Mark one of the most serious problems in science today.
    – barbecue
    May 13, 2023 at 18:23
4

This happens all the time (in my field, econometrics, mainly applied). I'll review papers for meta-analysis and two completely different authors will have done very very very very similar work, almost down to the specifications, and yet they came to these decisions independently. So I would cite the other authors, highlight your differences, and talk about how your method might be useful for your applications.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .