Names in the ACL Anthology

How the Anthology deals with names

January 15, 2026

This page explains how the Anthology creates author pages and groups papers by authors. Its main topics include:

Names in paper metadata

When a volume is ingested, we receive a list of authors for each paper as text metadata: given name(s) and family name, plus optional additional information. These “name strings” are what you see on PDFs and in BibTeX exports. (The Anthology policy is that these name strings should match what is on the PDF, which we treat as authoritative.)

We distinguish between a name or name string (the textual representation of a person’s name on a paper) and a person (the real-world individual behind the name).

Our goal is to create author pages which correspond to a real person. The difficulty is that a single person may publish under multiple name variants (diacritics, middle initials, different transliterations, name changes), and multiple people may publish under the same name. Both scenarios can be resolved either manually or automatically, as described below.

Name slugs

To create consistent URLs and to support matching across small formatting differences, the Anthology derives a name slug from a name. A slug is a normalized, URL-safe representation of a name: lowercased and hyphenated, without any punctuation or diacritics, other than a hyphen separator.

Each name slug has the format {given-names}-{family-name}, where spaces are replaced by hyphens.

Person IDs

When we manually resolve an ambiguous names to an author, we add a disambiguating third part to the name slug. By convention, this is the name or acronym of the institution where the author earned (or is expected to earn) their highest degree at the time they became known to the Anthology. For example, if there are two authors named “Alex Smith”, one with a Ph.D. from Stanford and the other first publishing while an undergrad at Tsinghua, their slugs would be alex-smith-stanford and alex-smith-tsinghua, respectively. These extended slugs function as unique person IDs.

Creating author pages

Author pages are created automatically when the Anthology site is built (following any change to the database). This is done by (a) reading names off paper metadata and (b) grouping them according to our internal database of verified authors.

A verified author is one for whom we have an explicit entry in our names database. This can happen either automatically (when a paper is ingested with ORCID iD information attached to the author) or manually (when we intervene to disambiguate authors with similar names or who publish under multiple names). We then create author pages of two types:

The presence of /unverified/ in the URL is a signal that the page was created automatically from name-only metadata.

For more on what “verified” means (and what the icons on author pages indicate), see Verification.

Correcting mistakes

When authors report problems (missing papers, wrong papers, split pages, or merged pages), we can:

Disambiguation or merging can be initiatied from the “Fix Author” button on each author page. This will fill out a GitHub issue template that we can use to track the request.

ORCID iDs: the best way to disambiguate

ORCID iDs are persistent identifiers for people. Providing an ORCID iD dramatically improves our ability to match papers to the correct author, especially as the community grows and name collisions become more common.

If you are an author, we strongly encourage you to create an ORCID iD and add it to your submission systems (especially OpenReview, and also Softconf where applicable). See ORCID iDs for best practices.