CONTEXT: In some domains (as mine), many research works end up with a prototype implementation of the research results. This prototype is usually released using some kind of open source license so that the community can benefit from it.
In AtlanMod , we are very proud of the large user base of our tools and of the benefits (and challenges!) that this brings to the team.
However, this has an undesired side effect, our tools are by far more popular than the research papers describing/introducing them (e.g. everybody in the software modeling community knows ATL ? but, how many could point to the paper that best describes ATL?).
If you are not a professional researcher you may be thinking “so what?”. The answer is easy: the current evaluation system for researchers (I´d say in any country) gives a lot of importance to the number of citations of paper but almost zero to the number of users of your tools. Simply put, a reference to our paper counts for the CV of all paper authors, indicating the url of a tool (even worse if it is just as a footnote or just by mentioning the tool) does not.
I’m not saying I agree with this (leading/creating widely used open-source projects should, and can, be quantified and evaluated as part of the achievements of a researcher) but until then please think about us the next time you use one of our tools in your research work. We´ll really appreciate it.
Btw, as Antonio Vallecillo proposed, to help on this, we should make clear ourselves which paper should be cited when using each tool.
Recently, this new GitHub feature is supposed to make things a bit easier. If you add a citation.cff file to your repo, GitHub will now suggest how to cite this repository based on the citation file contents. In theory, you can reference a research paper as the citation for the repo (as we have just done on our GH Xatkit repo) but the generated citation text will not look good. So, still work to do but a good step forward.
Hi, I think of course we should cite the paper that will contribute to the authors CV, but the author should make clear which paper we should cite, probably by promoting the paper in the tool website. One tool is possibly to have published several papers, and its hard to decide which one is “the paper that best describes ” the tool.
In addition it is possible that the paper was outdated comparing to the tool website. For example, xtext has a paper published in 2006, but it does not include the advanced features of the current version. So which one I should cite?
Qinan,
You’re right. This is exactly what I mention in the last paragraph of the post.
Jordi