Net­work sci­ence post­doc­toral research asso­ciates at North­eastern Uni­ver­sity have devel­oped a novel approach to iden­ti­fying com­mu­ni­ties in com­plex net­works, including major bio­log­ical net­works and large-​​scale social networks.

The results of the study were reported in a June issue of Nature magazine.

The team’s research find­ings could even­tu­ally be applied to solving com­plex prob­lems in fields of study as diverse as net­work sci­ence and mol­e­c­ular biology, says coau­thor Sune Lehmann, a post­doc­toral research asso­ciate in the Col­lege of Com­puter and Infor­ma­tion Sci­ence. Yong-​​Yeol Ahn and James Bagrow, post­doc­toral research asso­ciates at Northeastern’s Center for Com­plex Net­work Research, col­lab­o­rated on the report.

Using an algo­rithm cre­ated by the research team, biol­o­gists could poten­tially gain a deeper under­standing of com­plex dis­eases, such as cancer, by uncov­ering the rela­tion­ships among par­tic­ular groups of genes; and net­work sci­en­tists could explain the rela­tion­ship between a Face­book user living in South Africa and a stranger passing through Boston.

This gives us a whole new per­spec­tive on how society is put together,” says Lehmann. “It can help us under­stand how our social fabric is woven together and how we interact with each other and with the world.”

The break­through hinged on redefining “com­mu­nity,” and the role of “nodes” — the basic unit of a net­work struc­ture, such as an indi­vidual in a Face­book network.

Net­works, says Lehmann, are made up of com­mu­ni­ties, or densely con­nected groups of nodes, with a hier­ar­chical struc­ture in which each member can have only one affil­i­a­tion. But many real-​​world net­works — of Face­book friends, protein-​​protein inter­ac­tions or mobile phone users — have com­mu­ni­ties in which there is per­va­sive overlap, where each and every node belongs to more than one com­mu­nity.
Tra­di­tional approaches toward under­standing sys­tems of inter­acting objects have focused on grouping nodes, a strategy that cannot easily shed light on the rela­tion­ships between over­lap­ping communities.

North­eastern researchers, on the other hand, cre­ated an algo­rithm that rede­fines com­mu­ni­ties as groups of links, a method that resolves the unusual orga­nizing prin­ci­ples of over­lap­ping com­mu­ni­ties and hier­archy, making it clear how each node relates to every other. Link com­mu­ni­ties are small groups that have many rela­tion­ships between only a few objects. In a social net­work, such as Face­book, a link com­mu­nity would be a tiny group of friends with many friend­ships between them.

People have been doing this for a long time, but we found a new way of looking at the problem,” says Bagrow, adding that the accu­racy and strength of their algo­rithm increases with the den­sity of a net­work. As time goes on, and more data becomes avail­able, net­work sci­en­tists will be able to learn more about the sub­jects in the net­works they study.