DOC: Add .zenodo.json for author association #1597
Conversation
|
After initialization, I will post to Discourse requesting contributors add their ORCID. |
|
@thewtex - in nipype we have some custom auto inclusion based on commits + management. https://github.com/nipy/nipype/blob/master/tools/update_zenodo.py just a note that no system of contributors for a large system is good enough for everyone ! |
|
@satra awesome! I'll try the script... |
|
This is gold, thanks @thewtex. A related question, does somebody know what is the criteria of Google Scholar to parse Zenodo? It seems that some DOI's are integrated, but not all of them. |
|
Awesome @thewtex ! Adding a note to CONTRIBUTING.md looks fair. I think Nibabel has already some comment about it in their developper's guide, but the website seems to be down now. To complete what @satra has said, a few related thoughts in an earlier issue: If the script to parse authors based on the commits+management is finally found to fit our needs, we may want to add it to the Utilities/Maintenance folder, and add a section or note to the Release.md document.
No idea @phcerdan. Sorry. |
Ah, sorry, it was Nypipe that had a complete web page and details about how to cite the toolkit and inviting developers to open a PR to modify the .zenodo.json file with their data when contributing for the first time: They also cross-reference the GitHub contributors section. At the end of the page, they also have a section dedicated to acknowledging funding parties: might be an inspiration for #454. |
hjmjohnson
force-pushed
the
zenodo-json
branch
from
de5aaf2 to
12ead3e
Compare
I added a script based on this and a version in nibabel @jhlegarreta referenced. Running this script is documented in the release process notes so .zenodo.json is automatically updated.
Done.
Done.
The source of this is not clear or consistent -- I will exclude them from the CI to avoid false positives. |
|
FYI: https://orcid.org/orcid-search/quick-search/?searchQuery=Luis%20Ibanez
| |
|
Just an idea, but for credit in the zendo file, should we have a "long-tail truncation" where at least N (2, 3, 10, 50) commits are necessary in order to be listed? I'm always concerned that a long tail listing dilutes the benefits of being listed. My knee jerk reaction would be some number that includes approximately the top 50 developers. Just an idea. |
I am in two minds about this. I think giving credit to all contributors, regardless of the number of commits, makes the community a more welcoming place. GitHub lists all contributors. and the list can be ordered in different ways, always showing the top contributors by number of commits or additions/deletions. And there are other sites, if I am not mistaken, that list the top contributors according to different criteria. Remote module developers would not be credited by the current script either. May be this is also controversial, since they would not actually appear as having made a commit to ITK itself. Not sure if when the remote makes into ITK proper, the remote's history also makes it into ITK. If not, I assume they would not get their information into zenodo. Sorry Hans, more questions than answers. |
I agree with Hans, long-tails dilutes the benefits indeed, NCommits>70(?), or NModules >= 1? |
Based on https://github.com/nipy/nipype/blob/8fe05afcd7c716b52044837c96749ef420154fbb/.zenodo.json This mechanism explicitly associates authors with their [ORCID](https://orcid.org/). This commit seeds the configuration. Additional authors should self-identify by adding their ORCID to this file.
``` git shortlog -nse ```
As discussed in InsightSoftwareConsortium#1597, require 10 or more commits for .zenodo.json addition. This truncated the long tail, and simplifies the contribution process for new contributors by avoid the need to add an ORCID iD and affiliation to .zenodo.json. An "ITK Community Members" creator is added to the end of the .zenodo.json list to acknowledge the contributors with < 10 commits.
In d9a3ede a requirement for 10 commits is added. I think 50 is too high -- at 10, a contributor is often also contributing code reviews. This may also be helpful in reducing the burden for new contributors by not need to worry about adding an affiliation and an ORCID iD to .zenodo.json.
This is ok -- remote modules have their own repository, which can have a Zenodo record. |
As discussed in InsightSoftwareConsortium#1597, require 10 or more commits for .zenodo.json addition. This truncated the long tail, and simplifies the contribution process for new contributors by avoid the need to add an ORCID iD and affiliation to .zenodo.json. An "ITK Community Members" creator is added to the end of the .zenodo.json list to acknowledge the contributors with < 10 commits.
|
Merging -- @hjmjohnson if you create a separate PR for the ORCID's found, the contributors can review them. |
Based on https://github.com/nipy/nipype/blob/8fe05afcd7c716b52044837c96749ef420154fbb/.zenodo.json
This mechanism explicitly associates authors with their
ORCID.
This commit seeds the configuration. Additional authors should
self-identify by adding their ORCID to this file.
h/t @satra
We may want to add a note to CONTRIBUTING.md after merged.