Add wiki page on how to handle author page requests

[weissenh]

With the surge of author page request and the long backlog of such and only a few being handled, I'd like to propose to add a wiki page documenting how author page requests should be handled. There are ones for handling attachments and corrections (meant revisions?), revisions and errata and metadata corrections already.

This would also help me personally to understand what I need to do in order to help with this backlog. Plus a clear process would help both the assigned person and the reviewer. A few points that I would like to see or that I'm still trying to figure out right now:

• Background:
  • Link to related wiki page describing author representation (now: Name Variants, soon: Author page plan?)
  • Link to issue templates/how requests reach us
• checklist to follow?
  • assign oneself
  • do I need to check for duplicate requests, requests from different users about the same page, any associated metadata correction request etc.?
  • how much and how should the user request be verified?
    • highest degree institution, papers listed as theirs, canonical name if given
    • sources for cross-check? orcid, openreview, google scholar, personal or university homepage etc.?
  • tips and common problems/pitfalls if any
  • any help by automation alongside the process? Helpful scripts etc.?
  • how PRs should look like: bulk or one per person, what info to provide for reviewer to make review easy
  • single out one author vs full disambiguation of the name
  • how to decide canonical name if not given (most often or most recently used? most elaborate one?)
• for reviewer: what is needed to review a PR? What needs to be checked

I'm still figuring out how the process works, so I'd like to write that page once I have a better understanding of it. The page also needs to reflect the new author representation once it takes effect and changes to the issue template that come along with it (see #6416 )

[weissenh]

@Azax4 would you like to share any insights from your past efforts handling author page requests?

@mjpost since you've reviewed author page requests in the past: what would make reviewing easier for you? how do you review PRs on author pages?

[mjpost]

More documentation is a good idea. Have you see user-pointed corrections page? We also have a wiki page on name variants, which is user-focused but somewhat technical.

@Azax4 You have a good process; can you document this on the wiki?

[Azax4]

Hi both, the steps I generally follow are as follows:

1. Assign it to myself
2. Check if the form has been filled out with all the relevant data. Ping the author of the request if needed.
3. Quite often the author needs to just use the "Fix Data" buttons on the papers website to resolve issues. I request the author to do that and close the issue
4. Create a branch author-page-first_name-last_name-affiliation essentially author-page-new_id in most cases
5. Make an entry into the name variants file if needed
6. I usually use the script I have uploaded at Uploaded add_author_id_by_year #6421 to add author ids. I used to go to the xml files and change them all individually but using the script has reduced the time needed drastically. The script is fairly simple, it goes through all papers in a given year and adds an id that it takes as input wherever the first and last names match. eg. python add_author_by_year.py abhinav-gupta-mila Abhinav Gupta 2020 2019
7. I usually have to do this for both the new author page and a general catch-all
8. I look at the author pages before I start working and determine if I need to edit values somewhere. This happens when multiple authors have papers in the same year. If there is no clash then that's the best case scenario.
9. I make sure that the results are correct (counting each tag gives a good idea)
10. Create the PR, assign yourself, link to the issue, add to the appropriate Projects and Milestones

[Azax4]

To answer @weissenh

• Background:
  • Link to related wiki page describing author representation (now: Name Variants, soon: Author page plan?)
  • Link to issue templates/how requests reach us
• checklist to follow?
  • assign oneself
  • do I need to check for duplicate requests, requests from different users about the same page, any associated metadata correction request etc.? I don't really check for duplicates, if there will be duplicates in the same day you'll tend to remember the name or else it's usually merged. Not exactly fool-proof but it's worked well enough so far.
  • how much and how should the user request be verified? Unless something looks very wrong, I trust the person. I do open their links to make sure but haven't had any issues with it yet.
    • highest degree institution, papers listed as theirs, canonical name if given
    • sources for cross-check? orcid, openreview, google scholar, personal or university homepage etc.?
  • tips and common problems/pitfalls if any Make sure there isn't an existing entry in name_variants before adding an id. Make sure you are changing the author-id tag if using the script. It's not very hard if the details are provided correctly.
  • any help by automation alongside the process? Helpful scripts etc.? The one in the previous comment and bin[add_author_id.py](https://github.com/acl-org/acl-anthology/blob/master/bin/add_author_id.py)
  • how PRs should look like: bulk or one per person, what info to provide for reviewer to make review easy 1 per person is how it usually goes. I really like to see the links to the papers look like https://aclanthology.org/2018.iwslt-1.4/ where the exact year and id is visible. This makes it much easier to double check that the right tag is with the right file. But then someone might prefer a different format so I'm not too picky about it.
  • single out one author vs full disambiguation of the name I stick to singling out 1 author, the one who makes the request. We can always add more people later if they need it.
  • how to decide canonical name if not given (most often or most recently used? most elaborate one?) We don't have any formal guidelines but it usually goes firstname-lastname-abbreviation_of_uni_highest_degree Short forms of the abbreviation are okay only when really popular (eg. MIT)
• for reviewer: what is needed to review a PR? What needs to be checked The reviewer should usually check that the number of tags are matching and open up a few papers randomly, the first and last paper to make sure that the right papers are in the right place. There is a preview which is usually created which makes verifying things very easy

[weissenh]

@Azax4 You have a good process; can you document this on the wiki?

As I've assigned myself to this issue, I've provided a first draft for the wiki page incorporating what Azax4 has written above.
Please feel free to edit: https://github.com/acl-org/acl-anthology/wiki/Processing-Author-Page-Requests

Will get back to it maybe on Wednesday.

Have you see user-pointed corrections page? We also have a wiki page on name variants, which is user-focused but somewhat technical.

Yes, I'm linking both in the new wiki page too. I guess the name variants page is used to describe how matching works , but I like a separate page on how problems with matching should be handled.
Btw I'd like to suggest to also add a third point in the section about splitting author pages: aside from ORCID and institution, people should also explicitly list the papers that belong to them as a list of anthology URLs. Not doing so greatly increases time needed on our end (also Google Scholar links or listing the complement of your papers is somewhat helpful, but not ideal)

[Azax4]

Looks good to me, just made a minor edit for the branch naming convention