Requirements for open science

  1. IP: a permanent licence to use and create derived work based on the IP of the thesis should be freely given to everyone (CC BY-SA or similar). If existing company IP is needed to do so, a licence for it should be given as well. The IP obtained through the student’s thesis work will not be patented, or otherwise locked down from use by others through legal means.
  2. Confidentiality: in principle, all information related to the thesis project will be public, although sensitive information can be kept confidential (by request on a case-by-case basis to the student coordinator?) as long as it does not affect the full reproducibility of the thesis. Embargoes are not possible, thus the thesis report is published in the TU Delft repository right after graduation.
  3. Code: the source code produced by the student needs to be made permanently available (on a public repository like GitLab or GitHub) under an open source licence. Exceptions:
    1. If code from a company is used (to build upon), then the company’s code doesn’t need to be made open, but all of the interactions with it need to be fully documented (eg, detailed descriptions of input/output of all of the functions used by the student).
    2. If part of a student’s code only marginally improves the existing code of the company (e.g., bug fixes, small optimisations, documentation), then that part of the code doesn’t need to be made open either, but that part of the code won’t be considered as a significant result of the thesis either.
  4. Thesis (report) and presentation: detailed-enough methodology, analysis of the work and results to evaluate and fully reproduce the student’s work should be included in all cases.
  5. Data: data used (input + output) for the project should be made open and available on a site that is publicly available. Exceptions:
    1. If the thesis’ aim is to analyse a privacy sensitive dataset, a justification should be provided to the thesis coordinator and included in the thesis.
    2. If the thesis’ aim is to develop a methodology based on a dataset that contains personal data or is otherwise privacy sensitive, a small anonymised (or fake) open dataset should be created.