What is an academic field? Using co-citation networks to map scientific literature over time

Co-citations in Social Networks and Network Science by Raffaele Vacca on Vimeo.

What does an academic field look like? There are many possible answers to this question, as many as there are ways to look at what an “academic field” is. A way to describe academic fields, one that is gaining increasing popularity (see, for example, here and here), is to look at co-cited academic literature. We did this for the field of Social Network Analysis (SNA) and Network Science, and produced the figures and video showed in this page. The video displays the evolution of SNA and Network Science, as academic fields, from the mid-1990s to 2013.

Co-cited literature is the body of academic articles and books that are cited together, that is, are talked about together by the same scientists. Natural and social scientists write articles and books to share their findings, make scientific claims, and ultimately advance their discipline. The result is a sort of ongoing conversation between scientists, a conversation whose participants talk about their work as well as about the work of other authors, by citing them. Citations are the mechanisms through which science advances cumulatively, with new science talking about and developing from the results of existing science. When two scientists are co-cited, that means that authors are talking about them together. These two scientists are being read and studied together, in the same classes, labs, groups and research teams. Co-cited scientists are producing a cohesive, related body of scientific literature, which other scientists tend to connect and combine.

All this can be visualized as a co-citation network, where authors are nodes and they are linked if they are co-cited. Communities of close authors in this network are connected areas of scientific literature, whereas authors who are far apart in the network represent literature that is studied by different people, schools and research traditions. The result is a picture, a sort of semantic map, of the things that the academe talks about in a certain field, and of the way they are connected to each other. The field can be a discipline such as cultural anthropology, a cross-disciplinary field such as network science, or a substantive topic such as the psychological effects of child adoption. If the co-citation network represents a specific topic, different network communities are often different disciplines that are all interested in the topic but rarely talk to each other.

We created a co-citation network for Social Network Analysis and Network Science. Rather than just producing a static picture of it, we generated a video that shows the evolution of the network from the mid-1990s to 2013. This work won the Outstanding Visualization Award at the 2014 European Conference on Social Networks and is available and reproducible on Github using the R programming language.

To capture the field of Social Network Analysis and Network Science we obtained all the publications in the Thomson Reuters Web of Science that mention “social networks” or “network science” in their title, keywords or abstract. This is a traditionally multi-disciplinary field, which cuts across different disciplines including sociology, anthropology, psychology, computer science, and physics. We used Web of Science categories to divide publications into two broad disciplines – the social sciences versus physics and computer science. We then created the co-citation network of authors cited in this field. Each node is an author, and two nodes are connected if they are recurrently cited together by other authors in the field.

In the visualization node color represents the discipline in which an author is cited: (1) Blue authors are cited by publications in the social sciences; (2) Grey authors are cited by papers in physics or computer science; (3) Red authors are cited by papers in both the disciplinary classes. Only the top-cited authors are kept in the network, and a co-citation tie is only drawn if the two linked authors are highly co-cited, meaning the number of times they are co-cited is in the 90th percentile or higher of the distribution. This threshold was set because any two authors can be co-cited a few times for many different reasons, for example by general reviews of the field. This creates a sort of random noise in the co-citation network, such that two authors may appear as co-cited even though they are not meaningfully related in the literature. On the other hand, by only keeping ties between authors who are highly co-cited, with a number of co-citations above the 90th percentile, we only link names who are regularly and systematically associated in the field, thus we are able to reveal the deep network structure of this semantic map.

The figures and the video display a number of interesting patterns in the field of social networks from 1996 to 2013. The field has a so-called “core-periphery structure”, with a single major core literature in sociology and organizational science. However, starting in the early 2000s, a second smaller core emerges, consisting of academic literature from other disciplines, namely physics and computer science. This reflects the transition of “social networks” from a typical social science topic to a research area of interest to physicists and computer scientists. Social network research in physics and computer science steadily grows over the 2000s and up to 2013, when a subgroup of co-cited authors in physics and computer science is clearly visible as opposed to the traditional social science group.

The expression “social networks” once essentially referred to the study of human interactions, typically in small groups, by anthropologists, sociologists and psychologists. In the early 2000s, a fundamental change happened in the field. Online social networks such as Facebook and Twitter broke through, the “Big Data” era began, and new scientific fields such as Web Science and Computational Social Science organized. Network models started to be used in the natural sciences too, including biology and medicine, and “Network Science” gained popularity as a scientific field in its own. Today, social networks and network science designate a broader and more interdisciplinary field of research that engages social scientists, physicists and computer scientists alike.


POSTED: February 2, 2015.

Publication Types: 
Publication Date: 
Monday, February 2, 2015
BEBR Division: