Works Overview
A work is any scholarly document: a journal article, conference paper, book or book chapter, dataset, dissertation, preprint, and more. There are over 320 million works in OpenAlex, with tens of thousands added every day. A work’s OpenAlex ID looks like W2741809807; fetch one at api.openalex.org/works/W2741809807.
Works are the central entity in OpenAlex — everything connects to them. Authors write works, sources publish them, institutions get credit for them, topics describe them, and funders pay for them. Works also carry by far the most information of any entity: all twenty-ish entity types are first-class citizens, but works are the firstest class.
About
Where works come from
OpenAlex aggregates and builds on many open data sources. The catalog was seeded by the Microsoft Academic Graph (MAG) — Microsoft Research’s effort to catalog the scholarly literature — whose final open dataset OpenAlex adopted when the project was discontinued in 2021. The other primary source is Crossref, the open DOI-registration agency. Alongside those, works are drawn from DataCite, PubMed, HAL, institutional and subject repositories, and more; see How it’s built for the full ingest picture.
How locations become works
OpenAlex starts from locations: copies of a work hosted at particular places. A preprint on arXiv is one location; the same paper published in Physical Review B is a second; the accepted manuscript in a university repository is a third. The job is to recognize that all three are one work. Two mechanisms do that, used together:
- Persistent identifiers. If a location carries a DOI, PubMed ID, or arXiv ID that OpenAlex already knows, it belongs to that work. This is the certain path: a shared PID is treated as proof.
- Fuzzy metadata matching. Locations with no shared PID — most repository copies, many preprints — go through a fuzzy matching algorithm built on title normalization plus author names, tuned over the years for precision and recall.
Whichever path matches, the location joins the existing work and enriches it rather than minting a new one. If nothing matches, the location seeds a new work of its own; the only locations that don’t are ones with neither a PID nor a usable title, which are dropped.
Once a work exists, OpenAlex tracks its own metadata — title, abstract (and often full text), publication date, type — and the connections that make it useful at scale: authors, institutional affiliations, the source it appeared in, topics, funders, and citations.
How works cite each other — references, citation counts, and FWCI — has its own page: Citations. How OpenAlex decides whether a work is free to read: Open access.
In the API
The Works endpoint is at api.openalex.org/works. Fetch a single work by ID — /works/W2741809807 — or a list, and filter, search, sort, group, and page over the attributes documented on the Attributes page. For the full list of endpoints see the endpoints index.
Last updated
View as Markdown