Difference between revisions of "2020Q3 Reports: Publication Chairs"

From Admin Wiki
Jump to navigation Jump to search
Line 18: Line 18:
 
== Problems ==
 
== Problems ==
  
* We were not informed that https://github.com/acl-org/acl-pub/ is hosting the most recent paper styles until November 2019, long after we could have used that. We also found out about a FAQ in February 2020 (https://docs.google.com/document/d/1D41tt7Zj0_xjbxEaNWT6Y9NUcoBu0Lkp08IwlwX155E/edit#heading=h.nu0qu1d3gvr6).
+
* We were not informed that https://github.com/acl-org/acl-pub/ is hosting the most recent paper styles until November 2019, long after we could have used that. We also found out about a FAQ in February 2020 (https://docs.google.com/document/d/1D41tt7Zj0_xjbxEaNWT6Y9NUcoBu0Lkp08IwlwX155E/edit#heading=h.nu0qu1d3gvr6), again, long after it would have been useful.
  
 
* When the templates were posted to the Overleaf gallery, people saw them before the official announcement about templates went out.
 
* When the templates were posted to the Overleaf gallery, people saw them before the official announcement about templates went out.

Revision as of 23:00, 15 July 2020

Summary

Changes from last year

  • In the author instructions (which are also the paper templates), we changed the order of many sections so that important information wasn't scattered so haphazardly across the document. We also made other improvements, like using the recommended citation style consistently throughout the LaTeX source, and separating out all the LaTeX-specific stuff into clearly marked sections.
  • We wrote scripts to help scan for mismatches between author lists in camera-ready PDFs and author lists in START.
  • We wrote scripts to help scan for incomplete copyright forms in START.
  • We performed a very thorough manual review of all camera-ready PDFs and asked more than 400 authors to fix the many margin and other formatting problems we discovered. We wrote scripts to take a manually-populated spreadsheet of these formatting problems and email the appropriate authors. (START's mail tool was insufficient for these purposes since it could not populate portions of an email from an external spreadsheet.)
  • We wrote scripts to handle the custom program required by ACL 2020, including merging schedules from the main conference, demos, and student research workshop, prefixing papers with [Demo], [SRW], [Short], [Long], [TACL], [CL], etc., and working around an ACLPUB bug that includes a paper twice in the proceedings if it is in the schedule twice (which all ACL 2020 papers were).
  • We gave the three main-conference volumes real names in the metadata (“main”, “demos”, and “tutorials”) since that was now possible with the new ACL Anthology format.

Problems

  • When the templates were posted to the Overleaf gallery, people saw them before the official announcement about templates went out.
  • Since the author instructions are only in the LaTeX/Word templates, as soon as authors being writing their papers, they have lost their copy of the author instructions.
  • Demos, SRW, and workshops wanted to use the main ACL 2020 camera ready submission pages, but no one had copied those over (and pub chairs did not have the power to do so).
  • Many authors were confused about the new guidelines for supplementary material (i.e., that in the camera-ready any non-code, non-data supplementary material should now be an appendix instead of a separate upload).
  • Many, many papers had text, figures, and tables in the margins. There was no easy way for the authors to see where the margin problems were because on START, only the publication chairs saw the draft proceedings with the margin lines.
  • There is a wide variety of ways people format authors, several of which are not officially blessed by the style files. This makes it difficult to write a script that will automatically check that author lists on PDFs match author lists in START.
  • A small but significant fraction of authors were not getting hyperlinked titles of *ACL publications in their reference sections, presumably because they hand-entered the references rather than taking advantage of the anthology.bib that we included.
  • The default templates for building the proceedings (both main conference and workshop proceedings) include some confusing no-longer-used stuff like DOIs.
  • ACLPUB deletes titles that start with # (e.g., that start with hashtags). We found no solution to this except to work around it by manually deleting the # and reintroducing it manually into the schedule.
  • There was a last-minute need to delete a SRW paper from the proceedings that violated ACL ethics. We ended up having to use Adobe to just edit the proceedings directly, as re-generating anything so late would have likely introduced many new problems thanks to all the manual steps.

Recommendations

  • Point all pub chairs to the instructions from the previous point in the email in which they are invited to pub chair.
  • Add explicit guidelines for what author name formats are allowed into the instructions for authors (there are no instructions on that topic now).
  • Ask authors to include non-software, non-data supplementary material as an appendix even in the original submission, so that there is no difference in how this is included between original submission and camera-ready submission. Instruct reviewers that they do not need to read appendices (as we already do for supplementary material).
  • Modify the submission templates so that they clearly show style violations. The top priority should be flagging margin violations, but any other violations we can flag should be flagged in some bright color so that they can't be ignored, and will be caught long before the camera-ready.
  • Synchronize the publicizing of the templates and the posting to the Overleaf Gallery. Be aware that it takes Overleaf a day or two to approve an addition to the gallery.
  • Make copying the camera ready submission pages from the main conference to workshops part of the standard START setup that ?program chairs? do.
  • Fix the bug in ACLPUB that causes papers that start with # to lose their titles in the schedule.
  • Use Microsoft Teams for virtual meetings (rather than Zoom) if one of the pub chairs is in China.
  • Following previous years, we collected LaTeX source files from authors, but we don't do anything with those. Someone should decide whether we should continue doing this or not.
  • Rather than generating an intermediate "cdrom" structure and having the ACL Anthology ingest that, modify ACLPUB to directly generate the ACL Anthology format.