2021Q1 Reports: Anthology Director

From Admin Wiki
Jump to navigation Jump to search

Assistant hiring status

Last year ACL approved for me to hire an assistant. I ran an ad through the ACL Portal January of 2021. I have been running the interview process myself. Here are some statistics:

  • 37 applications (12+ female)
  • Selected 14 for half-hour video interviews (7 female)
  • Selected 8 of those for a final-round problem set (3 female)

The test is currently underway, and I will make a selection the week of March 21.

We also have about $3k from IAMT for ingesting the MT Archive website (mt-archive.info). After paying someone to digitize it, it now needs to be QA'd and ingested. I had been doing this myself but it is time-consuming. I will likely ask another one of the applicants to do this job.

Revision woes

We are spending a lot of time processing corrections to author lists, and I think the ACL should reconsider its policy a bit.

Many ACL conferences have adopted a form of the following policy that prevents any changes to the author list after submission (emphasis mine)

The author list for submissions should include all (and only) individuals who made substantial contributions to the work presented. Each author listed on a submission to ACL-IJCNLP 2021 will be notified of submissions, revisions and the final decision. **No changes to the order or composition of authorship may be made to submissions to ACL-IJCNLP 2021 after the abstract submission deadline**, which is January 25, 2011.

The Anthology treats PDFs as authoritative, meaning that we automatically correct metadata (such as names and titles) so as to be in line with the PDF. However, this could allow the ACL policy above to be trivially subverted: authors could submit their final PDF with an altered author list, and then request correction after ingestion into the Anthology. In order to prevent this, we require permission from the program chairs of a conference or workshop in order to make any changes to the author list, even when the metadata differs from the PDF. This has added some work. We receive requests along these lines perhaps every month. Most of them are requests are not to add authors, but for minor shufflings of author order among non-first authors.

The ACL might consider loosening this policy a bit. For example, you might allow author reorderings only (or, say, reorderings among non-first authors only). This might reasonably accommodate real-world differences in effort levels that arise after submission, without leaving open the abuse which I assume inspired this policy in the first place.

Style files and documentation

Until recently, there was no central repository for style files. Each conference would typically grab the style files from its previous conference. This meant that changes and bugfixes often didn’t make it across conferences (say from EMNLP to ACL), even within a year.

David Chiang and I made revisions to the style files and consolidated documentation in an attempt to address this. We have been in touch with conference organizers, and the files are being used for NAACL 2021 and EMNLP 2021 (we were too late in contacting ACL). The centralized documentation can be found here:

   https://acl-org.github.io/ACLPUB/

With style files here:

   https://github.com/acl-org/ACLPUB/tree/master/templates/latex

A working Overleaf template is available here:

   https://www.overleaf.com/read/crtcwgxzjskr

Here is a summary of changes:

  • Replaces \aclfinalcopy with a package option
  • Bugfixes related to the hyperref package
  • Pulled generic formatting information to the web page, which makes it easier to link to.

Up next

Next quarter, I plan to have updates on copyright and the status of video ingestion.