2026Q3 Reports: Anthology Director
Overview
The Anthology staff comprises Matt Post (director since 2019), a paid assistant (David Stap), and a number of volunteers. Over the past six months, the main active include myself, Nathan Schneider, and Marcel Bollmann.
A. Seza Doğruöz continues to handle submission of main conference materials to various indexing services (Web of Science, Scopus).
Nathan Schneider has organized biweekly meetings. This has been hugely helpful in keeping things moving, particularly ingestions. He has also putting large numbers of hours in helping address manual merges and splits of author pages under our new ORCID-iD-based system, and helping me improve documentation for interfacing with authors and contributors.
I have continued doing ingestions and much of the daily work myself, even without a paid assistant. Claude has been hugely helpful in offloading much of this work such that I think I'm fine without the assistant. Being involved with this lower-level work myself has a secondary benefit of keeping more me informed of new and ongoing issues.
Action item: making ORCID iDs mandatory
As listed in the Q1 report, we moved to a new author system based on ORCID iDs. This helps us automatically assign papers to authors, and supports the Anthology's goal of delivering reliable citable information. It also supports ARR in assigning papers to reviewers, since ARR reviewers are asked to link to their author page.
ACL main conference proceedings had [over 20,000 distinct authors], with over 9,000 of them new to the Anthology.
We have been active in alerting new ingestions to this policy, which helps keep our volunteer work down. ACL ingestion this year was held up a little bit by complications pertaining to conversion scripts that dropped exported iDs. Open Review exports them, as does Softconf. However, occasionally chairs do some of their own offline post-processing. We are working to improve messaging and documentation here.
- We would like to make submission of author ORCID iDs mandatory** for the ARR 2027 submissions, at least for ACL main conferences and workshops. We believe this would not be hard to do with cooperation from Open Review and Softconf. We have been in touch with Softconf about this and have support from Rich Gerber to do so.
Minor policy issue
Occasionally, an accepted paper will not be updated, and appears on the Anthology in its review form ([example]). Our policy so far has been to silently replace them if we hear from the authors, and otherwise to fully retract them, if we cannot get ahold of the authors.
Of note: hallucinated references
We have seen a lot of paper revision requests for papers with hallucinated citations. A big surge of these were filed after [the "hallucitation matters" paper] was posted on the arXiv. Thomas Dietterich [describes arXiv policy] that bans submissions from people found to have incontrovertibly participated in this.