Date Published: April 30, 2019
Publisher: Public Library of Science
Author(s): Jordan D. Dworkin, Russell T. Shinohara, Danielle S. Bassett, Naoki Masuda.
Scientific research is often thought of as being conducted by individuals and small teams striving for disciplinary advances. Yet as a whole, this endeavor more closely resembles a complex and integrated system of people, papers, and ideas. Studies of co-authorship and citation networks have revealed important structural properties of researchers and articles, but currently the structure of scientific ideas themselves is not well understood. In this study, we posit that topic networks may be a useful framework for revealing the nature of conceptual relationships. Using this framework, we map the landscape of interconnected research topics covered in the multidisciplinary journal PNAS since 2000, constructing networks in which nodes represent topics of study and edges give the extent to which topics occur in the same papers. The network displays small-world architecture, characterized by regions of dense local connectivity with sparse connectivity between them. In this network, dense local connectivity additionally gives rise to distinct clusters of related topics. Yet notably, these clusters tend not to align with assigned article classifications, and instead contain topics from various disciplines. Using a temporal graph, we find that small-worldness has increased over time, suggesting growing efficiency and integration of ideas. Finally, we define two measures of interdisciplinarity, one of which is found to be positively associated with PNAS’s impact factor. Broadly, this work suggests that complex and dynamic patterns of knowledge emerge from scientific research, and that structures reflecting intellectual integration may be beneficial for obtaining scientific insight.
The practice of scientific research represents the collective effort of humans to acquire information, generate insight, and disseminate knowledge. Although scientific inquiry has been carried out for centuries, the recent expansion of meta-data collection has allowed a robust body of literature to develop around the scientific study of science itself. This work has led to advances in predicting the success of scientific papers and authors [1, 2], found that articles often do not fit into existing disciplinary boundaries [3, 4], and provided empirical fuel for the debate over interdisciplinary research [5–8]. Yet much remains unknown about the nature of the large-scale scientific system that emerges from individuals’ intellectual and social incentives. It is especially unclear what features of this system may make it more or less effective at producing insights.
For this study, we used data from 65,290 articles published in PNAS between 2000 and 2017 to create a network of research topics. Though limited in scope, the choice to apply this framework to data from a single multidisciplinary journal was made for two critical reasons. First, the regularity with which disciplinary classifications are applied to articles sometimes varies across journals, and each journal has its own set of disciplinary classifications. The use of data from one journals facilitates a consistent and standardized system of classifications, allowing for investigation into the extent to which research topics do or do not cross disciplinary boundaries. Second, as the external relevance of the topic network was of interest, a single journal was desirable in order to draw connections between network structure and journal impact factor over time.
Prior analyses of collaboration and citation networks have produced deep insights into the structures and relationships behind the production of scientific research [9, 11–14]. Yet little is known about the network structure of the scientific ideas themselves, or what features of this network might be most effective at facilitating innovation. Here, we sought to present a generalizable framework for understanding the structure that emerges from relationships between scientific topics. By applying this method to data from PNAS, we demonstrate the value of this framework for characterizing the structure of research topic networks, investigating whether topic communities tend to fit into disciplinary classifications, quantifying how the landscape of topics is changing over time, and determining whether a network’s interdisciplinarity may be related to the amount of engagement that its component research receives.
In this study, we propose a topic network framework for investigating the emergent relational characteristics of concepts in scientific research, and apply it to articles published in PNAS since the year 2000. The topic network displayed small-world properties and interesting positive strength-betweenness/negative degree-betweenness associations, indicating the presence of tightly connected clusters and low-degree, high-strength nodes serving as conceptual bridges. Community detection showed that assigned classifications map poorly onto the underlying clusters, with a data-driven partition revealing the existence of multidisciplinary modules that contained topics from a variety of classifications. By investigating the temporal properties of the network, we found that both strength and small-worldness have been increasing over time. Interestingly, a novel measure of network interdisciplinarity was found to be positively associated with journal impact factor. Overall, this work demonstrates the value of network analysis in gaining insight into the structure of scientific knowledge, paints a picture of the surprisingly integrated nature of scientific ideas, and reveals a potentially important positive relationship between interdisciplinarity and scientific engagement.