OpenReview translator imports incorrect publication year for accepted ICLR papers

edited today at 7:16am
Hi Zotero Team and Translator Maintainers,

I’ve noticed a small but consistent issue with the OpenReview translator when saving metadata for accepted conference papers (specifically ICLR).

When importing an accepted paper from OpenReview (e.g., a paper published in ICLR 2026), the "Date" field in Zotero is often saved as the previous year (e.g., 2025).

This happens because OpenReview keeps the original metadata from the anonymous submission phase (which starts in the fall of the preceding year). The translator seems to grab this early submission/review date rather than the actual presentation/publication date of the accepted conference proceedings.

Since ICLR occurs early in the year, almost all accepted papers end up with a publication year that is off by minus one year in our Zotero libraries, which leads to incorrect citations (e.g., citing a 2026 paper as 2025).

Could the OpenReview translator be updated to check if a paper has been accepted/published in a specific conference edition (e.g., "ICLR 2026") and prioritize that venue year for the `Date` field, or parse the final publication date instead of the initial submission date?

Thank you so much for your hard work on keeping the translators so reliable!
  • We need an actual example URL, please, not the placeholder inserted by AI
  • edited today at 7:12am
    Any accepted paper from the ICLR 2026 Conference suffers from this problem (https://openreview.net/group?id=ICLR.cc/2026/Conference#tab-accept-oral).

    Take the first one as an example: saving this paper (https://openreview.net/forum?id=0wSlFpMsGb) through Zotero Connector results in the incorrect metadata "Date: 2025/10/08", which is the original submission date, not the actual acceptance/publication year. The paper belongs to ICLR 2026 instead of 2025.

    My apologies for the placeholder! I used AI to help structure my English description as a non-native speaker, but the issue itself is very real and reproducible.
  • Yes, understood, re:AI. Thanks for the example, I can reproduce that.

    Zotero doesn't have dedicated import from OpenReview. It just uses the generic metadata from the page, which, apart from the publication date is quite good. It wouldn't be hard to add dedicated support which could then easily fix the date issue (using the conference year or the published date, e.g.), but not sure how quick that'll happen. PRs welcome.
Sign In or Register to comment.