Network science: The language of integrative research

Please note: this commentary, recovered on 9-Jan-2017, was originally published in Science Dialogues on 16-May-2015.

Mathematics, they say, is the language of science. When it comes to what is happening—or has happened—down here on earth, it is beginning to look like the right dialect of mathematics to learn is what is now being called (somewhat confusingly) network science.

When the goal is integrating research discoveries across disciplines as diverse as archaeology, primatology, neurobiology, and geochemistry, the mathematics of networks is the Esperanto of choice.

Field Museum in Chicago is one of the world’s largest natural history and anthropology museums. Scientists working there study the world and its human inhabitants from scores of different research directions, both pure and applied. Integrating these often seemingly disparate specialities so that the results of so much scholarship can be communicated to the public through exhibits and publications has always been a problem.

Under the leadership of Thorsten Lumbsch, Ph.D., the Director of Integrative Research at the Museum, “The Field” as it is affectionately known in Chicago is pushing back against research specialization using network science. Here is one example.

A social network is a set of actors defined by their ties, links, or relationships with one another (e.g., friendship networks, ecological networks, global trade networks, and protein interaction networks) rather than by their individual characteristics (attributes) as actors. Since the research focus is on relationships rather than on characteristics, statistical methods  in network science are being developed that do not need to assume—unlike in traditional statistical analyses—that the observations being studied are independent of one another.

Dr. Termeh Shafie, who is currently a Visiting Bass Scholar at the Field,  arrived in mid April from the Algorithimics Department at the University of Konstanz in Germany to help the Field’s scientists apply the statistical methods and models of network science to their research datasets which are as seemingly dissimilar as gorilla social interactions, sharks swimming in the ocean, the genetics of lichens, and the decorations on prehistoric American potshards.

When asked about her work at the Field Museum in Chicago, Termeh Shafie explains:

"The first step will be to learn more about the empirical data at hand, the hypotheses about these data being considered, and how to embed a network approach to them. The second step will be to develop network models based on these hypotheses. This requires the mathematical formulation of models, programming these models using statistical software, and then running simulations. Goodness-of-fit tests can be used to test the fit of the models to the data. Once suitable models are identified, statistics can be used to measure different properties of the networks under study and unlock information in them using the models as predictive tools. Within a level of certainty, we can then predict trends and behavior patterns even for parts of the networks we don’t yet have data for."

On Wednesday, May 13th, Dr. John P. Hart (Director, Research & Collections Division, New York State Museum), Dr. Mark Golitko (Regenstein Research Scientist), and James Zimmer-Dauphinee (2015 Regenstein Intern) participated with Shafie in a small-group Network Science Workshop at the Museum exploring ways to apply network analysis to a large database of information about pottery designs on ancient vessels from 102 archaeological sites to help unravel how communities across southern Ontario coalesced between ca. A.D. 1350 and 1650 into the larger regional populations that ultimately became the historically documented Huron confederacy.

Left to right: John Hart, Termeh Shafie, James Zimmer-Dauphine, Mark Golitko
Left to right: John Hart, Termeh Shafie, James Zimmer-Dauphinee, Mark Golitko

Shafie will be at the Field until August 15th, but even after she returns to Germany, she will continue to be the “networks link” between scientists at the Museum and the Algorithmics Unit under the direction of Professor Ulrik Brandes in the Department of Computer & Information Science at the University of Konstanz.