Digital Object Identifiers (DOIs) for Electronic References

Electronic references are widely used in scientific publications. Most people are familiar with the URL (uniform resource locator), the address for a page on the Internet. A serious problem with URLs is described by the slang term link rot, the tendency of URLs to fail over time because the content has been deleted or moved. Markwell and Brooks1 found that science education URLs went bad at a rate of more than 10% per year (63% were bad after 5 years 11 months).

The DOI was created in part to address this problem. At the time of publication, DOI names are assigned to content objects themselves; in contrast, URLs are assigned to locations of objects. DOIs are permanent; once assigned, they cannot be changed. Because of the DOI system’s technology, DOI links are persistent—an object’s DOI link continues to work even if its URL changes.

All DOI names have a prefix and a suffix, separated by a forward slash; for example, 10.1001/jama.2009.2. All DOI prefixes begin with the directory code 10, followed by a period and a number identifying the entity that registered the DOI (usually the publisher), in this case the American Medical Association. The DOI suffix is a unique identifier for the content object and is assigned by the entity that registers the DOI; the suffix does not have to have any meaning, although some publishers use the object’s bibliographic information or its ISBN. In this case the suffix comprises the journal abbreviation (jama), the year of editorial processing (2009), and a sequential number (2). By convention, when a DOI is cited, “doi:” precedes the DOI and is set closed up to it.

The process by which a DOI is used to link to a content object is called resolution. There are 2 easy ways to resolve DOIs:

• Any DOI can be converted to a URL by adding the prefix “http://dx.doi.org/”; for example, http://dx.doi.org/10.1001/jama.2009.2. Enter the URL in any Web browser’s address window.

• The International DOI Foundation provides a Web page that resolves DOIs, http://dx.doi.org/. Paste the DOI into the box on this Web page.

To find the correct DOI for an electronic reference:

• The DOI registration agency CrossRef’s simple text query page (http://www.crossref.org/SimpleTextQuery/) locates DOIs for references input as text. Paste the reference into the box on this Web page.

• CrossRef’s fielded search page (http://www.crossref.org/guestquery/) supports a more targeted search.

• Use the click-through link from the PubMed service of the National Library of Medicine (http://www.ncbi.nlm.nih.gov/pubmed/). There’s often a link at the top right of an article’s citation page on PubMed to the article on the publisher’s Web site, and the DOI usually appears there.

See §3.15, Electronic References in the AMA Manual of Style (pp 64-67 in print) for more information about the correct format for DOIs in electronic references.—Paul Frank

1. Markwell J, Brooks DW. Broken links: just how rapidly do science education hyperlinks go extinct? http://www-class.unl.edu/biochem/url/broken_links.html . Updated July 24, 2006. Accessed January 17, 2012.

2 thoughts on “Digital Object Identifiers (DOIs) for Electronic References

  1. “At the time of publication, DOI names are assigned to content objects themselves; in contrast, URLs are assigned to locations of objects. DOIs are permanent; once assigned, they cannot be changed.”

    Like the entire DOI system, this is a bit breezy. The missing qualification is that DOIs are not unique, and thus “content object” doesn’t mean anything as operationally defined. Suppose one issues a commemorative issue of a journal consisting mainly of reprints, and recall that DOIs were originally touted as the solution to deep linking (something that never even came close to occurring). Does the reprinted version of Figure 1 of Article 1 have the same DOI as the original? Does Article 1 itself have the same DOI as the original? By necessity, no. The framework is just a label printer. As such, “because of the DOI system’s technology, DOI links are persistent” obscures the fact that there is *no new technology*–it’s merely a subscriber-based contractual arrangement in which one pays $35,000 a year for the privilege of adding an extra step to solving the original problem, which is keeping track of where you’ve put the data, and getting in return an uninformative user name and somebody else’s branding.

Leave a Reply

Your email address will not be published. Required fields are marked *