OpenAlex

Last month, we announced that we’re launching a replacement for Microsoft Academic Graph (MAG) this December–just before MAG itself will be discontinued . We’ve heard from a lot of current MAG users since then. All of them have offered their support and encouragement (which we really appreciate), and all have also all been curious to learn more. So: here’s more! It’s a snapshot of what we know right now. As the project progresses, we’ll have more details to share, keeping everyone as up-to-date as we can. As mentioned in our initial announcement, OpenAlex will be missing some data that MAG currently has–particularly at our launch in 2021, due to the very tight timeline. More accurately, we’ll have this data, but won’t be keeping it up to date. Specifically we won’t have: Conference Series and Conference Instances. Importantly, we’ll continue to bring in the vast majority of conference papers. But won’t be keeping track of new conferences themselves (eg, The 34th Annual Conference of Foo), and with that the ability to link conference papers to those conferences. Citation Contexts (the full text of the paragraph where each citation originally appeared) Most abstracts. We will however probably have those (minority of) abstracts that publishers send Crossref or PubMed for redistribution. Full coverage of DOI-unassigned works:. MAG is particularly good at finding scholarly papers without a DOI. We’ll be less good, especially at first. We will include many DOI-unassigned works…just not as many as MAG. There is some other data that may or may not make it into OpenAlex by December 2021. We are still testing these for feasibility: Although we’ll be missing some data, we’ll also be bringing some new data to the party — stuff MAG doesn’t have right now. Specifically will include: Over the long term, our goal with OpenAlex is to create a truly comprehensive map of the global scholarly conversation, so we’ll be continually looking to expand and enhance the data it includes. There will be (at least) two ways to get at the data: data dumps, and the API (below). Data dumps will be in the same table/column format as the MAG data, so that the downloads can be a drop-in replacement. There may be some additional tables and additional columns for new data we’re adding, and some data values will be missing (both of these are described above), but if you’re running code to ingest MAG dumps right now, you’ll be able to run pretty much the same code to ingest OpenAlex dumps in December. That’s a really important part of this project for us, because we know it will save a lot of folks a lot of time. The other way main to get at the data will be via the API. Here we will be doing it pretty differently than Microsoft. We will not be supporting the Microsoft Academic Knowledge API or Microsoft Academic Knowledge Exploration Service (MAKES). Instead, we will host a central, open REST API that anyone can query. This API will have two kinds of endpoints: entity endpoints, and slice-and-dice endpoints. Both will be read-only (GET), deliver data in JSON format, and be rate-limited but with high rate-limits. We appreciate that having a scheduled beta (or alpha!) release of the API and data dump would be very helpful. And we further realize that the sooner we can let you know that schedule, the better. Unfortunately, we don’t know the timeline for these releases yet. Our current best guess is the early fall. We’ll certainly be doing our best to get something pushed out there as soon as possible. We encourage you to join the mailing list so we can keep you up to date.