6

I have conducted experiment that had unexpected result which goes against consensus in my community and I worry whether it will be accepted by other researchers. For that reason I would like to make all my work public - the resources, the programs, the results.. Is there online service where I can publish the data for free and obtain DOI number that I can reference in my paper?

It makes over 2GB when packed.

I have heard that such services exist for people in biology and medical research but I work on transportation systems which is rather different (we do not produce so much data usually). I plan to submit the paper to journal published by IEEE.

ff524
  • 108,934
  • 49
  • 421
  • 474
student
  • 163
  • 3
  • 1
    Have you asked IT in your department or institution? Surely this won't be the viral download of the year...it's a large payload but given the small number of people who will want to d/l it, it may not be unreasonable to host it on an FTP server at your institution. – J... Apr 26 '16 at 22:41
  • PLOS One provides a long list of data sharing options which makes me think this is a big list type shopping question. – StrongBad Apr 27 '16 at 01:53
  • "I have conducted experiment that had unexpected result which goes against consensus in my community". Uploading the data or not, you should make a valid case in your paper, why your results make sense when they contradict the related community. No reviewer will have the chance to test the source and datasets while reviewing. So, you should probably worry more why the results were unexpected (bug?, weird test case?) and how to explain the results theoretically before anyone has the chance to check your source code... – Alexandros Apr 27 '16 at 04:13
  • 3
    @Alexandros As I read the question, it's his experimental data that goes against the consensus, in which case releasing it would make sense. In any case, my personal preference (and yours may be different) is to flip the burden of proof and ask why not to release the data relevant to your paper - it makes your work reproducible by others and is as such good scientific practice. – malexmave Apr 27 '16 at 07:26
  • 1
    @Alexandros I am not an autority in my field and I didn't imply my results are necesarilly correct. I tried to kill it, but failed to find the problem. So, I want to publish my work and my data for others to dissect it and/or reproduce it. Isn't this how science should be done? – student Apr 27 '16 at 08:31
  • This question has been asked and answered on OpenData SE. – gerrit Apr 27 '16 at 10:16

2 Answers2

13

You could also use Zenodo, which I personally prefer over Figshare since it is an open platform run by a scientific institution (the CERN) instead of a for-profit (compare the privacy policies of zenodo and figshare to get an idea on why this matters). They also guarantee that, should they ever shut down, they will migrate your data to another platform (and until then, your data is on the same infrastructure as the data created by the Large Hadron Collider (LHC)).

Their size limit is 2 GB per file, and you can add multiple files to a single dataset upload (which means you may not even need to pack your data, if it can be sensibly split into multiple files of <2 GB - and if you need more space, you can contact them). Finally, you also get a DOI for your data (which will be updated to point to the new home of your data if they ever shut down). Right now, it can be used free of charge while they develop a sustainability plan (later, they may charge you for uploading larger datasets, but there will always be a free tier with reasonable sizes).

malexmave
  • 1,226
  • 12
  • 22
6

Figshare:

  • Files up to 5GB
  • 20 GB private space, unlimited space for public data
  • Get a DOI
ff524
  • 108,934
  • 49
  • 421
  • 474
  • Thank you. However, it doesn't come with guarantee that the data will not be deleted once published. From their terms and conditions: "Company will not be liable for any errors or omissions in any content, and may delete, modify, or reformat any materials, content or information Submitted by you." – student Apr 26 '16 at 21:48
  • @student That's a standard TOS clause; I highly doubt you'll find any web publishing service that doesn't reserve the right to delete content they are hosting. – ff524 Apr 26 '16 at 21:51
  • You do make a good point but notice that it defeats the purpose of such service in a first place: I want to publish it there so the data are permanently accessible in the same way the paper is.. – student Apr 26 '16 at 21:59
  • 3
    @student It is permanently accessible exactly in the same way your paper is. IEEE also reserves the right to delete any content on its site in its TOS. – ff524 Apr 26 '16 at 22:00
  • 3
    @student You can't guarantee that anything is permanent. The best that you can do is to have redundant distributed copies of the data. I also encourage having a "home" for the project that you control, such as a domain with a simple web page, then you can update the link to the data if it ever changes in the future. – Austin Henley Apr 26 '16 at 22:07
  • Why not upload it in the mentioned service and write in your paper that in case of problem during downloading the data, they can contact you under your email (maybe one that doensn't expire as soon as you leave the institute...so maybe your private mail) and you can upload the data again and tell those who contacted you, that all should be fine again. Of course you'd need to have all saved on some (external) hard disk. – Lucas Apr 26 '16 at 23:13
  • There's a difference between putting it in a library or other archival system who budget for and plan for long-term storage and archival and a for-profit dot-com. I'm not sure which one FigShare is. – mako Apr 26 '16 at 23:23
  • What about dropbox and split the data such that they fit into the free storage capacity (2GB) of two accounts? – Lucas Apr 26 '16 at 23:34
  • @Lucas Dropbox limits total downloaded volume per day. Not the best option. – svavil Apr 26 '16 at 23:45
  • I added another answer for a service that guarantees to migrate the data if it ever shuts down - that's about as good a guarantee for permanence as you will realistically find, although they probably also reserve the right to delete your stuff should you upload anything illegal. – malexmave Apr 27 '16 at 07:20
  • @student: One of the key things is to get a DOI: service providers who assign DOIs to datasets have a contractual obligation to DataCite, who administer the DOI system for data, to keep the data available in the long term. So if figshare or Zenodo closed, they would have to transfer the data to another host or pay damages of some kind — that's not a cast-iron guarantee but probably the best you'll get. – Jez Apr 27 '16 at 09:29
  • @benjamin: figshare is a private company, part-owned by Palgrave Macmillan (the publishing giant), though see my previous comment about DOIs too. Worth at least asking your institution's library if they have advice, as most offer some support these days. – Jez Apr 27 '16 at 09:31
  • See this link for an explanation of how figshare attempts to ensure permanent archiving: https://figshare.com/blog/Ensuring%20persistence%20on%20figshare/25 – Jeromy Anglim Apr 27 '16 at 11:51