10

I wish to know whether there is some way, or maybe even some standard, for making a GitHub repository citable in a way that is recognized by academics. Many options have been suggested here, e.g., FigShare and Zenodo, but it seems to me that these routes are not actually established or taken seriously by academics.

For example, I made a Figshare DOI for my GitHub program in 2013 and it turned out to get zero citations, despite the paper related to the code getting many citations. People simply found citing something from Figshare to be awkward so the GitHub repository did not get citations the way an article would.

My question is this: Is there any practical evidence that some way of making a GitHub repository citable is somewhat accepted? For example, can anyone please point me to any GitHub repositories that have actually been cited many times as recognized by Google Scholar?

  • 3
    @aeismail: Unfortunately I am not interested in the answer to the question "how do I get people to cite my online repository". Honestly I think you're extending the concept of a shopping question to far outside of what it's original purpose was. I asked this question because I think there might be no examples at all, which makes this question very different from all the types of questions listed in the meta post that you gave a link to. How about if I edit the question to say "Is there any github repository with over 50 citations on Google Scholar without being published anywhere else?" – user1271772 May 27 '18 at 5:17
  • 2
    Then the question you should consider asking is: “How do I find examples of highly cited (code) repositories?” That gets you to your goal and still fits within our guidelines. – aeismail May 27 '18 at 6:39
  • 2
    @aeismail: I had read the meta link before making my comment 3 comments ago. I may be new here, but I have read the meta link you gave and took time to understand what a shopping question is and why they are unwelcome. All the reasons for rejecting shopping questions do not apply to this question. (1) The answers will not be based on opinion, they will be based on facts (link to Google Scholar page showing citations for a GitHub repo). (2) There is an objective way of deciding the best answer: the one with the most examples or examples with most citations. (3) It won't attract a lot of answers – user1271772 May 27 '18 at 18:35
  • 2
    @aeismail: Once again, I am not interested in how to find such things, and honestly I'm refraining from asking it that way because I am very doubtful that anyone will be able to suggest a strategy for how to find such a thing. I believe it would be easier to get an answer to the question I asked. This is because if someone just "happens to know" a highly cited Git repo, they can mention it (without having any "method" for how to find such a thing). In this sense, my question is much more likely to get an answer since any answer to your version will also include an answer to my version. – user1271772 May 27 '18 at 19:08
  • 2
    @Wrzlprmft: Well, I do see a point in having the list that I asked for. It would give me indication that there is some method, or maybe even some standard, for making a GitHub repository citable in a way that is recognized by academics. Figshare and Zenodo do not seem to be taken seriously by academics, but I might be wrong. I have no way to know whether or not I am wrong without this list. – user1271772 May 28 '18 at 0:40
9

The Keras repository has an entry in Google Scholar with 1254 citations.

I'm not aware of the author having done anything special to make it "citable" though, other than giving an example in the FAQ of how to cite it.


For example, I made a Figshare DOI for my GitHub program in 2013 and it turned out to get zero citations, despite the paper related to the code getting many citations.

Personally, whenever I wish to "cite a repository", the first thing I do is check the repository's homepage / documentation / FAQ / etc. to see if the author(s) wrote any instructions on how to cite it. If there are any such instructions, I use those since I assume that that is how the author(s) prefer to be cited.

If there are no such instructions, my next step is to see if there is any (peer-reviewed) paper published by the author(s) of the repository that describes / uses / links to the repository, and cite that if there exists such a thing.

Only if there is no such paper do I consider directly citing the github link itself. The implicit assumption is that, unless there are explicit instructions stating otherwise, it is preferable to cite (peer-reviewed) papers over repositories.

| improve this answer | |
  • If I try a new search and type Keras, the link with 1254 citations is to a PDF (that happens to be a dead link). – aeismail May 28 '18 at 13:04
  • @aeismail Yeah, I clicked the "All 4 versions" button for that one to get my link. That pdf appears to have been (old?) Chinese documentation (see how it starts with "KerasCN Documentation"), not a paper. The author's google scholar profile is here. He does have other publications, but none of them are specifically about Keras. – Dennis Soemers May 28 '18 at 13:16
  • But how would you know it’s about a repository? – aeismail May 28 '18 at 13:23
  • 2
    @aeismail Most people include the github url in the References when citing it. See, for example, this paper: www.jmlr.org/proceedings/papers/v48/gal16.pdf (the first one in the "Cited By" list on google scholar of the Keras entry) – Dennis Soemers May 28 '18 at 13:30
  • The Github link constitutes only 53 of those 1254 citations. – user1271772 May 28 '18 at 15:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.