Institutional Repositories

Introduction to Institutional Repository (IR) Interoperability

arXiv places no restrictions on whether articles also appear in local institutional repositories. Authors are welcome to download copies of their own articles from arXiv in order to submit to a local repository. This page describes ways in which institutional repository managers may approach finding and copying local researchers' content from arXiv.

Copying content from arXiv to an IR

Some institutions require or request that copies of articles written by their researchers are deposited in their local institutional repository in addition to arXiv. Everything necessary to pull complete metadata and fulltext from arXiv is available. However, the usual sticking point is the permission required to copy the fulltext into the institutional repository: arXiv does not have the right to grant such permission so in the general case permission must be obtained from the article authors. Obtaining permission from the article authors may not be necessary if:

there is a license permitting such copying associated with the article. The default arXiv license simply grants arXiv the right to distribute the article but does not authorize reposting in another repository. Licenses such as the Creative Commons Attribution license (CC BY) or the Public Domain Dedication do permit such reposting (see arXiv License Information for information about licenses supported).
there is some local rule or law that permits copying local researchers' articles into an institutional repository.

Procedure

We will consider the article arXiv:1410.6579 as an example.

Step 1 - Get metadata

Metadata from arXiv is available via our OAI-PMH interface, the URI for different metadata formats is constructed based on the article identifier. For example, to get oai_dc metadata the request is:

http://export.arxiv.org/oai2?verb=GetRecord&identifier=oai:arXiv.org:1410.6579&metadataPrefix=oai_dc

or to get arXiv format metadata, which has the license information expressed as a URI, the request is:

http://export.arxiv.org/oai2?verb=GetRecord&identifier=oai:arXiv.org:1410.6579&metadataPrefix=arXiv

Step 2 - Check the license

In the case of arXiv:1410.6579 the license is the Creative Commons Public Domain Dedication which is represented in the arXiv format metadata as:

...
  <license>http://creativecommons.org/licenses/publicdomain/</license>
...

The Public Domain Dedication allows and article to be copied to another repository without the need to ask for permission. Most submissions to arXiv use the default license however, expressed with the URI:

...
  <license>http://arxiv.org/licenses/nonexclusive-distrib/1.0/</license>
...

In these cases it is necessary to obtain permission from the article authors before the article may be copied to another repository.

Step 3 - Copy the PDF and/or source files

In the case of arXiv:1410.6579 the submission was in PDF format and the URI to download it is:

PDF: http://arxiv.org/pdf/1410.6579

In all cases links to the processed and the source files (where the submission is in TeX format) are provided on the normal abstract page (e.g. arXiv:1306.1073), they may also be constructed from the article identifier.

PDF: http://arxiv.org/pdf/1306.1073
Source: http://arxiv.org/src/1306.1073

See arXiv identifier scheme - information for interacting services and Media types delivered by arXiv for further technical details.

If you want to download just a few articles then there should be no problem provided a useful User-Agent string is sent in the HTTP requests, or if requests are made manually through a normal web browser. If you would like to download a significant number of articles then accesses should be spaced by at least 3 seconds to avoid our denial-of-service attack detector cutting off access, please contact arXiv support if you intend to download more than a thousand articles.

Identifying articles by your institution's researchers

Unfortunately, most arXiv articles do not have any affiliation information included by submitters, and when it is present there is wide variation in the writing of institution names which makes matching difficult. However, arXiv does maintain authority records linking articles to author accounts. This linkage is automatic for the submitting author but co-authors must claim-ownership after announcement in order to be linked. Additionally, user accounts may be linked with ORCID iDs and then a public display of all arXiv articles linked to that ORCID iD is available on arXiv in both human an machine-readable forms. With these linkages in place, if you know the ORCID iDs of your institutions' researchers it is then possible to find all their articles on arXiv.

The ability to link arXiv accounts with ORCID iDs was introduced in early 2015 and we suggest that institutions interested in identifying articles by their researchers encourage both claiming article ownership and ORCID iD linkage.

Example

Consider the article arXiv:1505.00009 which was submitted by first author Jonathan Heckman. Ownership was later claimed by co-author David R. Morrison who has also associated his ORCID iD with his arXiv account. If staff at UC Santa Barbara (UCSB), where David R. Morrison is faculty, wanted to find papers on arXiv but UCSB researchers they could query based on ORCID iDs. David's ORCID iD is http://orcid.org/0000-0001-6286-1277 and one can query arXiv using a URI of the form http://arxiv.org/a/ORCID, putting either the full URI or just the 16-digit part of the ORCID iD in place of ORCID, e.g.:

http://arxiv.org/a/http://orcid.org/0000-0001-6286-1277

or

http://arxiv.org/a/0000-0001-6286-1277

If accessed in a web browser these URIs return HTML pages. It is possible to request a machine-readable form either by explicitly appending .atom or .atom2 (see Author Identifiers for details of the two Atom formats), e.g.

http://arxiv.org/a/http://orcid.org/0000-0001-6286-1277.atom2

or using HTTP content negotiation with the header Accept: application/atom+xml, e.g.

$ curl -L --header "Accept: application/atom+xml" http://arxiv.org/a/0000-0001-6286-1277
<?xml version="1.0" encoding="UTF-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
  <title>David R. Morrison's articles on arXiv</title>
  <link rel="describes" href="http://orcid.org/0000-0001-6286-1277"/>
  <updated>2015-09-23T00:00:00-04:00</updated>
  <id>http://arxiv.org/a/morrison_d_1</id>
  <link href="http://arxiv.org/a/morrison_d_1.atom2" rel="self" type="application/atom+xml"/>
  <link rel="describes" href="http://arxiv.org/a/morrison_d_1"/>
  <entry>
    <id>http://arxiv.org/abs/1507.05965v2</id>
    <updated>2015-09-23T08:31:55-04:00</updated>
    <published>2015-07-21T16:00:43-04:00</published>
    <title>On Gauge Enhancement and Singular Limits in $G_2$ Compactifications of M-theory</title>
    ...
  </entry>
  ...
</feed>

An attempt to request information for an ORCID that does not exist or is not linked to an arXiv account will result in an HTTP 404 Not Found response, e.g.:

http://arxiv.org/a/http://orcid.org/0123-0123-0123-0123.atom