2021Q3 Reports: Anthology Director
Things are going well with the Anthology.
Quick updates
Ordered by decreasing suspected average general appeal to the ACL Exec.
- We continue to have problems managing and storing videos. We do not have existing files dating back to NAACL 2013 in order, and there is no consistent, enforceable policy for new videos. We continue to work on these two issues, but do not have much bandwidth for this.
- As of July 1, the official canonical page for the ACL Anthology is https://aclanthology.org/. Old links from the previous site at https://aclweb.org/anthology/ still work (via 301 redirects). We plan to use this page until the end of time. We are also working to see whether the LDC is willing to redirect first-generation Anthology links from https://acl.ldc.upenn.edu/.
- We have [a new copyright form], which I hope to have live by EMNLP 2021.
- I hired an assistant, Xinru Yan, who has been working with me since April, handling day-to-day operations. This has been immensely helpful, and I don't know how I did this work without her. She works about 5--8 hours per week. We easily have enough work to employ another person at this level, if there are funds for it.
- We are ingesting the MT Archive, which contains proceedings of Machine Translation conferences back to the 60s. This is being done by David Stap (a Ph.D. student in MT) with funding from the IAMT. This should be completed by September.
Requiring Exec attention: retractions
We have had to process a handful of retracted papers, which requires us to adopt a policy for this, as well as make decisions for how to disseminate and display them. Here is our proposed policy, which I have built with input from a number of people. Please note that I plan to implement this policy, so now is the time to speak up if you have comments or suggestions. The policy below was crafted to balance two competing interests: the need to disseminate retractions, without unduly punishing authors for their honesty.
- We follow [The ACM policy on retractions].
- The paper is processed as a revision. A prominent WITHDRAWN watermark is added to each page of the paper. A note at the top directs the reader to the paper page in the Anthology.
- The paper page itself displays a prominent error message. The title, author list, and abstract are stricken. No bibtex is generated.
- In the volume listing, the paper appears with its title and author list in stricken text.
- On the author page listing, the paper is not displayed at all.
The idea behind excluding the paper from the author listing page, and that page alone, is to avoid creating a permanent, prominent black mark on the author's record. In all other settings, the retraction is prominently noted.
https://preview.aclanthology.org/retractions/
Future plans
We are working with next year's conference planning teams (specifically, Dan Roth and NAACL 2022) to rewrite the ACLPUB tool, with the goal of simplifying the publications and ingestion process. More detail will follow from appropriate venues in the months to come.