Basic DOI Syntax. A Digital Object Identifier, or DOI, is a unique string created to identify a piece of intellectual property in an online environment. Detailed information about the DOI itself can be found in the DOI Handbook at http://www.doi.org/handbook_2000/index.html. For the purpose of assigning DOIs and registering them with CrossRef, you need only familiarize yourself with the contents of this brief document.

The DOI is made up of two components, a prefix and a suffix, separated by a forward slash. See figure 1.

All DOI prefixes begin with "10" to distinguish the DOI from other implementations of the Handle System.

How to get a DOI prefix. Your unique six-digit DOI prefix will be assigned to your organization by CrossRef and is covered by your CrossRef membership fee. It is therefore not necessary to get a DOI prefix prior to joining CrossRef or to pay a separate fee to the International DOI Foundation. However, please do review the IDF’s terms and conditions for DOI prefix at http://www.doi.org/terms.html.

It is worth noting that while specific publishers assign DOIs using a unique prefix, once a DOI is assigned it can be transferred to another publisher if a journal is sold. Therefore, the prefix of a DOI does not reliably identify the publisher of an article.

In order to assign DOIs to your articles or other content items, you must come up with a DOI assignment scheme. Since the DOI prefix is provided to you, the next step is the DOI suffix.

The DOI suffix for a given journal article or content item must be created and assigned by the publisher or other organization with legal authority to register the entity in question. The DOI suffix has a very flexible syntax. It may be any alphanumeric string of your choosing, consisting either of a single node or multiple nodes.

Because the DOI is an opaque string (or "dumb number") intended to remain unique and persistent throughout changes in copyright ownership and location of the content entity, no specific or descriptive information needs to be given in the DOI. Rather, such information forms the metadata associated with each DOI. If bibliographic information is included in a DOI string, it will have no meaning within the CrossRef or DOI system. Schemes based on primary keys that are unique and meaningless are also less likely to run into an exception that violates the scheme than those based on bibliographic information.

In addition to the uniqueness requirement, we recommend you employ the following guidelines in creating DOI suffixes for your content entities:

1. The suffix should be as concise as possible within reason, in consideration of human readability. DOIs will be displayed online and in print and will be re-typed by end users.
2. It should reflect a consistent, logical system that can be easily documented and readily understood by employees of your organization, so that the task of assigning DOIs can be passed from one employee to the next. You might therefore want it to include existing internal identifiers in use within your organization.
3. It should not be based on print bibliographic metadata (such as volume, issue, page), both because it should allow for the assignment of DOIs to online articles that are published in lieu of or ahead of print, and because the number is explicitly not intended to facilitate generation of identifiers by third parties.
4. Consider URI Syntax (http://www.ietf.org/rfc/rfc2396.txt), and please refer to the Unicode 2.0 character set (www.unicode.org). Some characters like ampersands and asterisks cause problems in URLs and need to be encoded. UTF-8, which preserves ASCII characters, is the required coding. Because SICIs include special characters, using them within a DOI can lead to difficulties in encoding. See the NISO DOI Syntax standard for more information - http://www.doi.org/handbook_2000/appendix_1.html
5. Suffix nodes may be used to reflect hierarchical information or levels of granularity. For instance, the first node might be a multiple—letter code for the journal title, while successive nodes encode year of article acceptance and order of article acceptance. This is the scheme used by Academic Press, with resulting DOIs like DOI:10.1006/jmbi.1998.2354. Similarly, in the case of electronic books, the AAP recommends the following syntactic structure for the DOI suffix:

   /Whole work. Next granular level. Next granular level. Etc.

6. DOI schemes should be extensible, and the suffix nodes may be used for this purpose. For instance, in the future, parts of articles such as figures, graphs, and supplementary materials will be assigned DOIs. Using the AP example above, the second figure in this article might be assigned the DOI: 10.1006/jmbi.1998.2354.f002.

Sample DOIs. Crossref publishers have established many varying schemes for creating their DOI suffixes. Here is a list of sample DOIs for some current CrossRef publishers:

Work vs. manifestation. DOIs can be assigned to any type of intellectual property in any medium. Detailed information about DOIs can be found in the DOI Handbook - http://dx.doi.org/10.1000/182. More specific information about works vs. manifestations can be found in chapter 9 of the Handbook. As a matter of policy the CrossRef DOI identifies the work, not its various potential manifestations. This means separate DOIs are not assigned to each format of a given article. The print, PDF, and HTML versions of the same article will all share the same DOI. Different versions of the same article can be pointed to from the response page to which the DOI resolves. In future with multiple resolution, more than one DOI will be registered with a DOI.

[Section on CrossRef as Registration Agency?]

CrossRef policies on DOI deposit. Please note that it is not necessary to register DOIs separately with the International DOI Foundation since this will be done automatically as part of the CrossRef metadata submission process. For technical details on the metadata submission process, go to http://www.crossref.org/docs/CrossRef_batch_spec.pdf.

A DOI for a given content item should be registered as soon as electronic full-text for that given content item comes online. While that DOI may have been created and assigned prior to publication, it is extremely important for the reliability of the CrossRef resolution system that such pre-publication DOIs remain for internal use only, and that they not be deposited or distributed to secondary publishers prior to full-text publication. At the same time, we require that DOIs be deposited as soon as possible after publication. (Elaborate?)

When metadata and DOIs are deposited with CrossRef, the publisher must have active response pages in place so that it is able to receive incoming links for those content items. As soon as metadata is deposited in CrossRef, other users of the system will be able to retrieve your DOIs and create reference links to your content. A minimal response page consists of a full bibliographic citation displayed to the user. The vast majority of CrossRef publishers also display abstracts for free. Authorized users should be authenticated for full-text access, and unauthorized users may be presented with pay-per-view options if available.

It is the publisher’s responsibility to maintain the accuracy of the metadata, DOIs, and URLs. There is no charge for sending updates or revisions to previously submitted records.

Every DOI is registered with a URL. At present one URL is registered for one DOI, but in future multiple URLs (and other data types) will be registered with each DOI. The core functionality of the DOI system is to resolve a DOI to the registered URL. For a detailed explanation of resolution by the DOI System and multiple resolution see the DOI Handbook - http://www.doi.org/handbook_2000/resolution.html.

The DOI Proxy server — http://dx.doi.org/ - resolves DOIs. To resolve a DOI, the DOI number itself should be attached to the address for the proxy server. For example, the DOI 10.1006/jmbi.1998.2354 would be made an active link as http://dx.doi.org/10.1006/jmbi.1998.2354. It is important to note that in this example of a URL, the DOI number 10.1006/jmbi.1998.2354 incorporated in a URL and transported by the HTTP protocol and therefore has to follow guidelines for URIs. The syntax for URIs (a URL is a URI) is much more restrictive than the syntax for DOIs. For information on the DOI Syntax and character issues see the NISO DOI Syntax standard - http://www.doi.org/handbook_2000/appendix_1.html.

In future HTTP may no longer be the dominant protocol and browsers like Internet Explorer will handle DOIs natively. This means that doi:10.1006/jmbi.1998.2354 would be picked up by the browser and automatically resolved. For more information on this, see section 6.5 of the DOI Handbook — "The resolution interface with Handle System technology" - http://www.doi.org/handbook_2000/resolution.html#6.5

A main purpose of DOIs in the CrossRef process is to create persistent links to full text content, especially in reference links. DOIs for references are retrieved from CrossRef and added to references lists. How the links are actually added depends on the publisher’s online system.

A DOI for a piece of content (article, book chapter, conference proceeding) should also be displayed in the header information for the online AND print versions of the content (see Figure 1). The DOI can also be used for citing the content. For example, many publishers instruct readers to use a DOI to cite an online article published without a volume, issue or page number.

A citation of the Science article with a DOI would be "Cell Proliferation Without Neurogenesis in Adult Primate Neocortex", Science, Volume 294, Number 5549, Issue of 7 Dec 2001, pp. 2127-2130, doi:10.1126/science.1065467

The Science article above has reference links. An example is reference 36:

Behind the display text of "CrossRef" is the URL http://www.sciencemag.org/cgi/external_ref?access_num=10.1038/35016083&link_type=DOI

This links sends the DOI for the reference through the local HighWire system that resolves outgoing links and passes them to the DOI Proxy (http://dx.doi.org/)

Instructions for Authors:

Publishers should include information about using DOIs in their instructions for authors.

[Examples?]

• DOIs, along with volume, issue and page numbers, should be part of the standard bibliographic metadata for an article
• DOIs should be displayed in bibliographic headers for online AND print articles
• In citations DOIs should be presented as doi:10.1038/35016083 (doi should be lowercase and no space should be between the doi: and the start of the DOI string)
• Wherever possible, the DOI should be an active link but the http://dx.doi.org/ need not be displayed — e.g. doi:10.1126/science.1065467
• In a reference link to full text, CrossRef recommends that "CrossRef" be used as the display text for the link. More information on this can be found in the "CrossRef Branding Guidelines"

---

Figure 2 - DOI in Bibliographic Header

Information for the end-user

If you encounter a DOI in text or header information, you can resolve it by embedding it in an http link to the DOI resolver dx.doi.org, which redirects the DOI to the currently registered location for this content item. For example, the DOI: 10.1006/jmbi.1998.2354 can be resolved as http://dx.doi.org/10.1006/jmbi.1998.2354 If you click on this link, you will arrive at the appropriate response for at the Journal of Molecular Biology.

There you will see that the DOI is included in article header information or on the title page. To include the DOI in a citation to this article, you simply append it at the end, prefaced by "DOI:" as follows:

Brian G. Turner, Michael F. Summers. "Structural Biology of HIV." Journal of Molecular Biology, 285(1), pp. 1—32. doi:10.1006/jmbi.1998.2354

Note that "doi" here may be either in caps or lower case, and may be preceded by a comma or a period, depending upon the stylistic conventions of the publication at hand. Because of the way the DOI system works, this DOI-based link will remain persistent over time, even if the location of the article moves or ownership of the journal changes hands.