6

I have performed some experiments on approximately 400+ data sets, my PI is thinking of publishing online a report that contains some background information, the software and the procedure used to obtain the results and a detailed table of the results and maybe some analysis and conclusions of the results.

Both the software used to do the tests and the data sets are publicly available, and we didn't contribute in establishing them, i.e. we didn't code the software, nor did we collect or produce the data sets.

He expects that this report will get cites in the future, at least by his group's future work.

My question is: What is the academic value of such report for a student who wants to pursue a phD?

  • I am confused by the term "report". Is he going to send this report on a journal (and make it a paper) or just want to upload it online? – Alexandros Jul 12 '15 at 19:03
  • "performed some experiments on [...] data sets" might bear explaining. I normally think of "data sets" as things whose existence can result from experiments (but not only from experiments), not as things on which experiments are done. – Michael Hardy Jul 12 '15 at 19:21
  • @Alexandros upload it online as he thinks it is not acceptable journal material. – The Hiary Jul 12 '15 at 19:25
  • 2
    It seems perfectly conceivable that whatever it was you did (but it's not clear what that was) could be a contribution to knowledge and not available from just the data sets and the software. Your bottom-line question seems impossible to answer without knowing the nature and customs of the academic field you're working in. There's an immense amount of variation from one field to another in such things. – Michael Hardy Jul 12 '15 at 19:25
  • 1
    @MichaelHardy Running experiments on standard data-sets is fairly standard practice in computer science, where the gathering and curation of datasets is often separated from the tasks people would like to do with them. Dataset collections allow better comparison of methodologies because people are using similar benchmarks. Some well-known examples include the UCI machine learning repository, and the TIMIT speech corpus. – jakebeal Jul 13 '15 at 4:48
3

Short form: if the contribution is scientifically useful, then it will also be useful for your career.

It seems to me that the implicit question within your question is whether you should be investing your energy into something else instead, like try to turn it into a peer-reviewed publication in a high-impact-factor journal, and whether this sort of less formal publication is a "waste of time." If your PI is well-experienced in journal publication and decides this isn't well-suited for a journal publication, then I would be inclined to trust your PI's judgement.

The nice thing about doing a more informal publication like this is that it can also potentially be much more lightweight to prepare and publish: you don't have to follow anybody else's formatting requirements, nor worry about what peer reviewers will attack you on. You can just write it up nice and simple and the way you'd like to and post it online.

Then, if it's useful to other scientists, it will be cited and you will get credit. Not as much as for a journal paper, but quite likely with a similar "value for effort" ratio for you. Moreover, computer science is a particularly friendly field for getting credit for such alternative forms of contribution, in which many scientists have become famous for their tech reports and software projects rather than their journal papers. And if it turns out not to be useful to other people, trying to publish it "higher" instead would likely be a worse use of your time in any case.

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.