Making a journal article findable
A published article reaches the readers it was written for only if they can find it. That depends on identifiers that keep the article and its authors attributable, on how readers browse journals and hunt for articles online, and on writing and linking the article so that search engines find and rank it, without letting optimisation outweigh good scientific practice.
Digital identifiers
Digital identifiers were introduced over the decade to 2020, with the growth of information technology, to identify digital objects uniquely and stably, and scientific publishing is one of their applications. Digital objects can be archived with their metadata to preserve them (the "memorable web"), made findable through deep web search (the "navigable web"), or given a unique, persistent identifier that also gives instant online access (the "identifiable web"). Using web addresses as identifiers carries risks: they may not persist, so that a server migration or a transfer produces "error 404", and one object can have several of them. The main solution proposed for an identifier that is unique and persists in the long term is a digital identifier in place of a web address. A digital identifier is a code unique to its object, like a fingerprint, and is issued by an independent authority, which enhances its reliability and accuracy. Clicking it leads, through a four-step mechanism called resolution that a browser completes in under a second, either to a valid web address for the object or to its metadata, such as its creator, publisher and format. Its main functions and benefits are unique and persistent identification, location information kept current when the object moves, standardised and enriched metadata, easier searching, copyright protection, more traffic, lower costs and time savings, new business and research areas, and increased revenues.
In scientific publishing, digital identifiers and the systems built on them can help those involved in writing, editing, submitting, reviewing and publishing papers, and an identifier acts as an article's fingerprint on the web, connecting it by hyperlink to millions of other articles. For scientific outputs such as articles and datasets, digital identifiers provide unique, persistent identification and persistent access at once.
An article identifier is a number the editor assigns as soon as the article is accepted, and it distinguishes that publication from every other. A preprint may already carry an identifier, but it is not the final one and differs from the one issued after peer-review approval. PubMed and MEDLINE assign a further unique number of their own, by which the article can always be found there.
Digital identifiers have specific applications in scientific publishing, among them these five:
- Unique and persistent identification. Identifiers create unique, persistent, automated links between journals, publishers, editors, authors and even an article's tables or datasets, removing human error in maintaining those links, and they enable automatic, reliable retrieval of information about articles.
- Acceptance by indexers. Many indexers, such as International Scientific Indexing, PubMed, Scopus and the Directory of Open Access Journals, ask in their initial evaluation of a journal whether it uses identifiers. This is not stated as a requirement for indexing, but journals that use them generally find acceptance much easier, and it has been claimed that a journal indexed by one of the major indexers that starts using them can see its impact factor rise greatly within a year.
- Automated citation monitoring. Identifiers can be used to monitor automatically the influence of an article, an author or a journal.
- Visibility. Information about articles with identifiers is recorded in multiple reputable databases, which raises their visibility in search engines. Bringing an article to as many readers as possible is a big challenge for every journal and author, and the added visibility benefits journals and authors and helps readers find articles.
- Plagiarism prevention. Identifiers help maintain copyright by giving direct access to an article's official links. The plagiarism detection service iThenticate includes in its database only published articles with identifiers and enters them automatically, and it notifies a subscribing journal and its publisher when another article with even the slightest resemblance to one of theirs is published.
The Digital Object Identifier
The Digital Object Identifier (DOI) is the best-known digital identifier system. It is managed and controlled by the International DOI Foundation, a non-profit organisation, and its registration agencies, ten in 2020, allocate DOIs on the foundation's behalf. The DOI system and its registration agencies, such as Crossref and DataCite, are among the most important applications of digital identifiers in scientific publishing. A DOI is allocated permanently to give a stable hyperlink that redirects to current information about its object, including its location; that information can change over time, but the DOI does not. A DOI is a character string of indefinite length made of a prefix and a suffix. The prefix is the directory code, always 10, followed after a full stop by the registrant code that a registration agency assigns to the owner or publisher. A forward slash separates the prefix from the suffix, an alphanumeric code that the registrant assigns, that must be unique to the registrant and that may include an ISBN, ISSN or serial number. There are no operational limits on the length of a DOI or its components, and identifiers up to 4 GB can be assigned. To resolve a DOI, https://doi.org/ must be placed in front of it.
Crossref, an independent organisation launched in 2000 and run by research publishers, exists to facilitate access to, citation of and reuse of articles and books as research outputs. Publishers of scientific content, especially books and journals, can use its services, including DOIs. Through DOIs it lets a reader switch from article to article at a click, even across journals. Its other DOI-based services include Similarity Check, developed by iThenticate to let publishers check the authenticity and originality of submitted articles; Crossmark, which shows whether the version being read is up to date; and Crossref Metadata Search, which automatically provides the information some indexers need without the journal's involvement.
Publishers in developing countries use DOIs significantly less than those in developed countries, often from lack of awareness and money, so their researchers' work is read and cited less, and the citations it does receive cannot be tracked. One author holds that this is the most important gap that wider use of DOIs could help reduce. A systematic study of articles indexed in PubMed found DOIs used more by journals in developed than in developing countries and by international than regional journals. To increase and improve sharing further, the study's authors called for measures to allocate DOIs to journals in developing countries. The same author expects that, through the use of identifiers in low- and middle-income countries, their research outputs will become more accessible to the world at large, narrowing the research gap between richer and poorer nations, and so suggests that readers, authors, reviewers and editors of research become more familiar with digital identifiers and the services built on them.
Author identifiers
Attribution is ascribing a piece of work to a particular person, and in academic institutions accurate and reliable attribution of research outputs such as publications is of prime importance. Common surnames can complicate it, as a 2013 paper whose authors are Yu, Wang, Li, Liu, Li, Wang, Wang and Wang shows. National censuses show a high likelihood that two unrelated people share a family name, and anyone called Johnson, Patel, Lee or Garcia can describe the mix-ups that such surnames often cause in private and professional life.
ORCID was launched in 2012 to give researchers a unique identifier and so address the problem of correctly attributing scientific publications. It is a machine-readable 16-digit digital identifier, similar to a barcode, that aims to link researchers with their academic output regardless of changes of affiliation or name. ORCID was not the first attempt to disambiguate authorship. Many publishers had for years issued log-in names or unique IDs to submitting authors, and the ISO-certified International Standard Name Identifier already covered contributors to creative works and those active in their distribution. ORCID is the academic equivalent of that standard, and the two have issued a joint statement on interoperation. ORCID claims uniqueness in reaching across disciplines, research sectors and national boundaries and in cooperating with other identifier systems. In 2013 it was supported by major players from different sectors, including universities, research organisations, scientific societies and publishers, and most publishers, including Wiley, Nature Publishing Group, Elsevier and Thomson Reuters, had integrated it into their electronic submission of peer-reviewed manuscripts.
In 2013 ORCID was used to track more than peer-reviewed publications. It also covered conference abstracts and posters, researcher contributions that may not qualify for authorship status, and even social media posts, and it could be linked to pre-existing publication identifiers such as Scopus Author ID and ResearcherID or to a LinkedIn account, with bibliographic databases including Ovid and Europe PubMed Central linked to it as well. Funders also took part: the US National Institutes of Health subscribed in January 2013 and began testing ORCID in its grant application workflows, and the Wellcome Trust was actively involved in its board. The Trust's head of evaluation expected that ORCID might become a gold standard in research and lead to new quantitative metrics for evaluating scholarship and scientific merit. Registration is free. In 2015 ORCID was streamlining research administration and might give insights into researcher productivity.
One practitioner suggests that every medical writer register with ORCID and CRediT, describing both as digital identifiers that distinguish individual researchers in publications and better acknowledge their contributions. By the same practitioner's account, ORCID is a simple website where any researcher can register, and in 2021 a registered identity opened the submission pages of a growing number of journals directly, marked the holder as a trusted author, researcher or medical writer, and made it easier to find the details an electronic submission system asks for when an article goes to a journal or a congress.
How clinicians read
One physician-author holds that clinicians do not read each journal cover to cover and proposes that they read in three ways: they graze, hunt and gorge. By his account, which he concedes rests on no real data, clinicians are most likely to graze the paper journals that cross their desk, in their spare time, because there are too many publications to read each in depth. As a subscriber, he increasingly receives previews of these journals online. Grazing, as he does it, means reading the table of contents carefully, turning to the abstract when a title looks interesting, and reading the methods, results and summary only when very interested; he estimates that he reads the abstract of every third or fourth article and about a quarter of those in depth. Several journals print brief synopses of their major articles, such as the NEJM's "This Week in the Journal" and JAMA's "In This Issue of JAMA", and these may be all that many busy clinicians read. He tends to give more attention to the two major broad-based journals, JAMA and the NEJM, and to the leading journal in his specialty, so less prestigious publications get less attention.
Hunting is how clinicians look for information about a specific problem, and it is now done online: the same author seldom goes to a paper journal to hunt for a clinical fact. Hunting makes it vital that important studies are published in indexed journals: those who read the work are likely never to hold the journal, but by locating it online and citing it they disseminate its findings. Gorging, reading every word of every article, is a luxury few readers have.
Search engine optimisation
Search engine optimisation is the process of optimising writing so that it can be found easily online. Once the domain of technology colleagues, it was still a relatively new concept in 2023 for most academics, researchers and medical writers, although many large publishers by then advised authors on optimising their articles before publication. All medical writers should now understand it and apply it actively to their work.
By 2012 students and researchers rarely visited the library and reached research articles almost exclusively online. The shift from print-driven to online journals requires small but important changes in writing style that raise an article's visibility and so the likelihood that it is downloaded and cited. After 40 years in print, the journal Politics & Policy, for example, entered 2012 as an online-only, subscription-access publication. Google and Google Scholar are the principal search engines by which people find an article online. The search engine is now a researcher's first port of call, and it is of paramount importance that an article can be found easily in its results. One publisher has disclosed that over half the traffic to its online library comes from Google, Google Scholar and other search engines, a share that highlights the importance of optimisation for the discoverability of research. Researchers rarely look beyond the first 20 results from Google. Research that is easy to find online is viewed, downloaded and cited more than a paper buried on page ten of the results. Good optimisation can increase a work's discoverability and impact, and in a rapidly growing body of scholarly literature it helps people find the work and lets it be shared with the broader community.
Web crawlers, spiders and bots find the pages and files that search engines list. Search engines rank results with complex, proprietary mathematical algorithms that weigh more than 100 factors without disclosing their weighting. The algorithms change every month to keep the results as accurate as possible. The elements they consider include the volume of incoming links from related websites, the volume and consistency of searches, time within the website, page titles, page views, quality of content, revisits, relevance, click-throughs, page descriptions, technical user features, quantity of content, uniqueness, technical precision of source code, key words, functional rather than broken hyperlinks, and spelling.
The starting point is the content the author writes. An author should write for the audience but be mindful of how search engines work too. Five key steps can be taken to increase discoverability: choosing the keywords carefully, optimising the title, optimising the abstract, using the keywords through the rest of the article, and building links to the work.
The body, figures and files
The keywords can also be used, judiciously, in the sub-headings, the article text, tables, figures, captions, image alternative text, graphics and the document's file name. Search engines can read and index the content of vector graphics, such as Encapsulated PostScript, Adobe Illustrator and Scalable Vector Graphics files, which can be made in PowerPoint. They do not index the content of other image types, though information can still be saved in those images' file names. A graphical abstract therefore needs thought about both its file format and, if it can be a vector graphic, the keywords it contains. Other outputs saved to the web should have keywords in their file names (File-Name-SEO.pdf, for example), and a PDF should have relevant metadata added to its file properties and be saved as an archivable PDF/A.
The main body should be structured with subheadings so that a reader can navigate by jumping to a section. A coloured graph is no longer enough to impress; videos, podcasts, sound files and animations have taken its place. Novel ways of representing data lead both to increased citation and to download by people looking for teaching tools, and resources and editing software for them are freely available.
Links to and from the article
Links both to and from an article can further improve its optimisation. Citing seminal articles and the authors' own relevant research builds links and increases discoverability, and listing every author's ORCID identifier connects the new paper to each author's past work. An article should be connected to the literature it sits in, and an engaging, comprehensive literature review can increase the chance of citation. A range of articles, books and online data sources should be cited, and the number of pages an article is connected to also features in search engines' ranking algorithms. Giving the web addresses of cited articles, with access dates where relevant, enables reference linking.
Search engines weigh the volume of links coming into an article: the more connections to it, the higher it ranks. One way to raise that volume after publication is to market the article by telling academic and social networks about it and sharing its web address on X, LinkedIn, Facebook, Instagram, a personal or company website or blog, Academia, Mendeley and ResearchGate. It can also be sent to colleagues and broadcast through the many modern communication and educational tools, such as Twitter, Moodle and Wikipedia.
By 2009 a reader could become a fan or follower of journals such as The New England Journal of Medicine, JAMA and Science on Facebook or Twitter. Links on the journals' Facebook pages took users to articles and to features such as audio commentaries and podcasts, while tweets drew followers to new articles of interest.
Some of the adaptations that online publishing asks of authors, particularly adopting social media, may seem daunting, but stepping back to consider and modify one's approach to publishing online will undoubtedly reap rewards.
The limits of optimisation
Black-hat optimisation, practices that raise a site's ranking by breaking the search engines' rules and policies, can lead larger search engines to penalise work by showing it much lower in the results. The form most relevant to medical writing is keyword stuffing, overusing keywords until they appear out of context. Repetition of key phrases should stay natural, because Google will un-index an article that goes overboard in repeating terms. Optimisation should be considered for every kind of medical writing that will be on the internet, journal articles and reports alike, but it should never outweigh good scientific practice and publication.