New Guidelines for DOI Linking and Display


For sometime I have been collecting digital object identifiers (DOIs) in my BibTeX entries, as described in this blog post. When I use my own BST file to format the references BibTeX creates hyperlinks to the published source via the DOI. If I use the the SIAM BST file the DOI is instead displayed as part of the reference.

The Crossref organization (which provides DOIs) has recently issued revised guidelines on the display of DOIs. Up until now, DOIs have typically been displayed as, for example, 10.1137/130920137, and linked to as The new guidelines say that the link should be and that it should always be displayed in this form, as a full URL. Note that the “dx” part of the URL has gone, and “https” has replaced “http”.

The main reason for the change is that the pure DOI on its own is not much use, as it can’t be clicked on or pasted into a browser address bar without first adding the prefix. Additionally, https provides more secure browsing than http, and Google gives a small ranking boost to sites that use https.

Crossmark states that the old form of URL will continue to work forever.

I have updated my BST file myplain2-doi.bst in this GitHub repository, which contains a BibTeX bibliography for all my outputs, so that it produces links in the required form.

SIAM has updated the BST file in its macro packages to implement the new guidelines.

PeerJ Computer Science: A New Publishing Experience

The download button and the button to (un)follow an article.

PeerJ Computer Science began operation in early 2015. I’ve just published a paper in the journal, and am also an editor of it. PeerJ Computer Science does a lot of things differently than journals that I’ve published with before, so I thought it would be useful to explain what is different and novel about it.

PeerJ Computer Science is an open access journal whose articles are published under a Creative Commons Attribution license. As with its older sister journal Peerj (life, biological and health sciences), authors pay a fee to publish. I won’t discuss the pricing details, which can be found here, but just note that many institutions already have a publishing plan with PeerJ, which means that authors at those institutions will find their articles are pre-paid. This was the case for me, though I had to pay a $99 fee for my co-author (a PhD student).

The editorial board is large: over 300 editors, each of whom has to choose which subjects, from a given list, match their research interests. There is no Editor in-Chief. When a paper is submitted, the PeerJ office contacts all editors whose interests match one of the subjects chosen by the submitting authors, asking if they would like to edit the paper. If no editor responds, after reminders if necessary, the editorial office regards the paper as either not in the scope of the journal or insufficiently interesting and returns it to the authors—though I believe this is not a common occurrence.

The journal’s aims and scope statement says

“PeerJ Computer Science only considers Research Articles … PeerJ Computer Science evaluates articles based only on an objective determination of scientific and methodological soundness, not on subjective determinations of ‘impact’ or ‘readership’ for example.”

In every other journal in which I’ve published, significance and likely impact are criteria for acceptance, so PeerJ Computer Science is very different in this regard. It’s too early for me to say how these easier-to-satisfy criteria affect the refereeing and editing process.

The coverage of the journal is defined by its list of categories. Closest to my interests are Algorithms and Analysis of Algorithms, Data Science, Optimization Theory and Computation, Scientific Computing and Simulation, and Software Engineering. Each of these categories forms an overlay, pulling out a subset of papers from the journal as a whole that lie in the given area.

The journal aims for a fast turnaround. Referees are given 2 weeks or 3 weeks (the editor chooses) to provide a report. PeerJ Computer Science gets a first decision back in about a month.

Here are some of the things I like about PeerJ Computer Science.

  • The process of submission, refereeing, and editing has been designed to be web-based, and it is very nice to use. The submitting author has a lot of information to complete, but much of it relates to the journal’s policies: funding sources, competing interests, and data availability must be entered, along with a description of what contribution each author made to the paper (something new for me, but standard in many areas of science).
  • I spend a lot of time on journal websites trying to find the “download PDF” button and the “export citations” button, which seem to be in a different place on every site. PeerJ has a delightfully simple big blue Download button (see the image at the top of the page): click it, and you can select what you want to download, be it PDF, BibTeX, or something else. What a brilliant idea!
  • All authors are given the opportunity to set up an author profile page, which provides a link to their PeerJ articles as well their website, Twitter account, GitHub account, ORCID, etc. See my profile and my co-author Weijian Zhang’s profile.
  • The journal has a thorough set of policies dealing with all aspects of ethics and procedures. These are laid out extremely clearly.
  • Once your paper is published you get access to a personal “To-do list” web page with ideas for how to publicize your paper, which include Tweeting about it and emailing colleagues, with one click producing a partially completed Tweet or email. The page explains “Why promote your work?” and records which To-dos have been done.
  • While the published paper is frozen, the page it sits on (pointed to by the DOI) is not. As an author you are able to add links at the end of the page, perhaps to blog posts, updated software, or follow-on work.
  • Referee reports and author responses can be made publicly available via a “Peer Review History” link, provided both authors and referees agree. The history for my paper is here. The original submission can also be downloaded.
  • The PeerJ staff read the reviews as they come in, and flag anything that might be problematic with the editor, such as an inadequate review. This is a great idea. I do enough editing, for various journals, that usually I look at reports only when they are all in. Early notification of issues with a review can shorten the time to a decision being made on a paper.
  • PeerJ uses LaTeX, and high quality PDF files can be downloaded. Papers are displayed in the browser in html form, with MathJaX used for the mathematics (see this example); they look very similar to the PDF version.
  • While PeerJ does not copy edit manuscripts, it does put them into the journal format and copy edit references into the journal style. The experience from my one published paper was very positive, and included a short email exchange about how best to format one reference. After having had a poor experience with copy editing and production at a commercial publisher recently, I found the PeerJ proofing stage a pleasure.
  • For readers, the web site works exactly as you would hope. Searches are fast and accurate. Thanks to the responsive web coding, papers can be read in html form comfortably on an iPhone.
  • The journal integrates with a preprint server, PeerJ Preprints, which supports versioning. Authors can start a submission at either PeerJ Preprints or PeerJ Computer Science and then export their submission to the other, retaining metadata. I have not tried PeerJ Preprints, but it offers an interesting alternative to the ArXiv or an institutional preprint server.
  • With its Paper Now experiment, PeerJ allows you to create, edit, and display a journal article entirely in GitHub. For an example see this GitHub article and this corresponding PeerJ article.

In summary, PeerJ Computer Science is completely different from all the journals I have previously published in or edited for, but I am impressed by what I’ve seen. By rethinking how a journal should be managed and published in the 21st century, the PeerJ team have brought some fresh ideas into this domain of academic publishing.