How to effectively manage citations

One of the cornerstones of the academic article is citations. Scholarly knowledge proceeds by recognising the work of others, which means by citing published articles and books, and then using that knowledge to create some new theory or idea that is different in kind to what came before. After the new paper has been published, others will critique it, might agree, or disagree with it, and so scientific knowledge continues to evolve. As Newton wrote, in a 1675 letter to Robert Hooke, “If I have seen further, it is by standing on the shoulders of giants.” In other words, he read all the relevant articles and books that preceded his own thinking.

Putting citations into context

It was Eugene Garfield in the 1960s who came up with the idea of the citation index, as a way of assessing the relative importance of scholarly articles. It was built on the idea that the more citations an article has, the more influential it has been: “the citation index … may help a historian to measure the influence of an article — that is, its ‘impact factor'”. Today, the impact factor is familiar to everyone in scholarly publishing, but we are equally familiar today with the limitations of citations. An article may be cited because it is supported, but it could be cited because the author thinks the cited article is wrong. An article may be cited because it uses a standard methodology that has become widely accepted, so the new article can be seen to be using a trusted and reliable base. Taking into account these nuances, a simple count of all citations is not a very helpful metric, and that’s before considering citations to articles that subsequently modified or retracted. There have been several initiatives to replace or to improve on citations as a measure of quality, but the traditional citation count persists.

Managing multiplying citations

Even if all citations are treated equally, the challenge is that they multiply the work of the researcher enormously. It wasn’t so difficult in Newton’s time, when the number of journals could be counted on one hand, but today, a typical scholarly article will have, say, 25 citations; each of the cited articles will in turn have a further 25 citations; and so on; and with over 25,000 journals publishing articles on a regular basis, that is a vast body of literature that has to be appraised before you can start writing.  Let’s say you are writing an article based on selection you have made of 50 published papers. To read all the cited articles from those papers will involve reading 25 x 50, or 1,250 papers. And to read all the papers cited by those would require reading 31,250 articles – clearly an impossibility.

However, there is now a new generation of citation tools available, for just this purpose. Semantic Scholar, Scite.ai and Scholarcy are three of the best known, but each has a different emphasis. Broadly speaking, scite.ai indicates if a citation simply mentions a paper, or if it confirms (agrees with) or contradicts the cited paper. Semantic Scholar has a tool to indicate influential citations; it also differentiates citations for methodology and citations for background.

Semantic Scholar indicates influential citations

Citations can be viewed in two directions: which articles cite this paper (a forward citation), and which articles are cited by this paper (a backward citation). Scholarcy can identify how a citation builds on, or differs from, previous work.

Scholarcy tool for comparing a paper with its predecessors: where it agrees, and where it disagrees

By using these three tools, it becomes feasible for researchers to sift through hundreds or even thousands of papers to identify the chain of significance in a research area, before having to read individual articles. Somewhere among those hundreds of papers, there are a few that were genuinely innovative, that changed the scholarly landscape. The real engagement with the researcher takes place, of course, when those key articles are read in detail; but using these tools can save many hours of time reading articles that are simply derivative.

The role of libraries in citation management

What is the role for libraries in all this? Many researchers will simply not be aware of the citation tools today available, and that some of these tools are complementary, rather than duplicating the same functionality. Once researchers have seen the benefits of these tools, they tend to adopt them enthusiastically, but how can they keep up with the latest developments in research technology when there are new productivity and analytics tools appearing every week? This is where the library comes in. Libraries provide a wide range of how-to guides to software that assists the scholarly workflow, for example a guide to using bibliographic software, or a page on using Google Scholar. Similar pages could help researchers make the most effective use of citations in the articles they read.

Newton’s admission that he was drawing on earlier researchers’ work was essentially the same process that takes place on a much larger scale today. Although, in a touching note, it turns out that Newton wasn’t the first person to use the analogy of standing on giants’ shoulders. According to Wikipedia (and who would doubt that source), the phrase was first used in the 12th century by John of Salisbury. Newton should really have cited his sources.