Announcing a New Affiliation Matching Strategy in the ROR API

We are delighted to announce the launch of a new matching strategy in the ROR API to help you match complex affiliation text strings to ROR IDs at scale.

Publishing and repository systems have traditionally stored author and affiliation information as unstructured data, capturing intermingled details about an organization as a single text string in a single field. This unstructured information makes it difficult to reliably connect research outputs with associated institutions.

Many repositories and publishers are beginning to use ROR instead, recognizing the importance of producing and sharing open, structured affiliation metadata, and here at ROR we continue to encourage Crossref and DataCite members in particular to include ROR IDs in DOI metadata. Often, a crucial step for Crossref and DataCite members in ROR adoption is to convert text strings extracted from submitted manuscripts or stored in legacy data to ROR IDs, and this is one reason the affiliation matching endpoint is the most-used endpoint of the ROR API.

We know that accurately matching text strings to ROR IDs is a priority for many users, and therefore we have been continuing to develop ways of meeting this need with our API and with other services.

How and why we developed the new affiliation matching strategy

ROR’s first affiliation matching strategy, developed by Crossref’s Director of Technology, Dominika Tkaczyk, was announced in 2019. It has been available in the ROR API ever since. This strategy works by breaking long affiliation strings into substrings and making multiple queries against the ROR registry. If a match is found, the ROR API returns the ROR record identified with a chosen:true indicator.

Beginning in 2023, the ROR team began collaborating with Crossref again to develop a new matching strategy that would be both faster and more precise at scale. Throughout 2024, the principles behind this collaborative research were explained in a popular series of blog posts about matching on the ROR blog co-written by ROR Product Manager Adam Buttrick and Dominika Tkaczyk.

Early in 2025, we reported that this collaboration between ROR and Crossref had produced a new, faster, more precise strategy for matching affiliation strings to ROR IDs. Unlike the original “multisearch” matching strategy, the new strategy performs a single search of the ROR registry to find the most likely match, and therefore we’ve dubbed it “single search.”

OpenAlex quickly incorporated this new strategy into its own affiliation matching, and Crossref used it to produce a dataset of 95 million affiliation strings from its metadata matched to ROR IDs. At ROR, we then began the work to include this strategy in our own API, which was recently completed and released last month. Like all of ROR’s code and data, the single search strategy is freely and openly available.

Benefits of the new affiliation matching strategy

In a demonstration of the single search strategy given at the November 2025 ROR Community Call, Adam Buttrick explains that the new strategy has been designed to reduce the rate of false positive matches, because we know that accuracy is of primary importance for unsupervised automated matching. The single search strategy is more computationally efficient, better able to find relevant results, and more accurate than the already high-performing multisearch strategy.

In just one of several examples drawn from real Crossref metadata, Adam shows in the below video that a string such as “Department of Cognitive Sciences, School of Social Sciences, University of California, Irvine, California;” is not matched by the current multisearch affiliation strategy on the ROR API but is correctly matched by the new single search strategy to the ROR ID for the University of California, Irvine: https://ror.org/04gyf1771.

Using the new affiliation matching strategy

To try out the new single search matching strategy, append the term &single_search to a ROR API ?affiliation query as described in our updated documentation.

Note that the single search strategy is currently in a preview period, so the existing multisearch matching strategy is still the default for ?affiliation queries. Both matching strategies will remain available in the ROR API for the foreseeable future, but we plan to make the single search strategy the default for ?affiliation queries in the first quarter of 2026.

To give us feedback, you can open an issue on our roadmap or contact us at support@ror.org. Please note that any individual string might be less well matched using the new strategy than using the existing strategy, so we are most interested in tests that use sizable open datasets.

ROR’s essential purpose is to improve metadata throughout the scholarly information ecosystem so that reliable connections can be made between organizations and the research they support. Clean, normalized, structured, open organization information helps ensure research integrity and enables better tracking and analysis of research. Our hope is that by constantly improving ROR’s systems and services we can improve scholarly metadata for everyone.