By Marlene Cimons, National Science Foundation
Online social networks such as Facebook or LinkedIn, already popular with millions for their easy ability to help people forge connections based on common interests, also have become ideal "laboratories" for social scientists who want to study human behavior.
"The volume of online social networking is exploding, and it appears it is becoming more pervasive than real-life social networking," said Dan Stefanescu, professor of mathematics and computer science at Suffolk University in Boston, where he is directing several projects using online social networks as a research tool. "This is not surprising, given the ease with which one can pursue interactions in the digital world."
Furthermore, in an online social network environment, "it is so much easier to find new 'friends' for social intercourse because, unlike in real life, one can reach the 'friends' of 'friends' of 'friends' of your 'friends,' ad infinitum," he added.
Data available from online networks can provide insights into certain behaviors—how people will vote in upcoming elections, for example, or what consumer products they are likely to buy—and also can prompt blueprints for engineering new social systems, and predicting certain events and economic outcomes, he said.
In the job market, for example, information obtained from social networks can improve "the ability to find better employer/employee matches which, in turn, imply better productivity, and reduction in wasteful jobs and hiring search costs," he said.
Furthermore, activity on massive online social networks potentially could help small businesses grow more quickly, via word-of-mouth advertising on Twitter, for example, or enhance the marketing of certain products that might not otherwise be heavily advertised, such as independent films.
The process also could create new businesses by transmitting ideas and disparate pieces of information among members. Stefanescu cited a recent Economist article, for example, that described how a Dutch technology consultant, frustrated by the short lifespan of his iPhone's battery, complained about it on LinkedIn, and was directed to a small company in China that produces small plug-in batteries for the phone. Impressed with the product, the consultant told members of his online community about it, and soon was handling orders for the Chinese company. Later, he formed his own business to manufacture these plug-in units, incorporating the Chinese firm's battery.
Conversely, information dissemination on massive online social networks "is of great interest to the advertising industry, which is interested in the impact of ad placement, i.e., viral marketing," Stefanescu said.
Moreover, collections of public postings also can serve as a "human sensor network',' even an early warning system, by confirming or predicting such events as power failures, outbreaks of infectious diseases, or street demonstrations, among other things.
Typically, research using online social networks involves collecting very large amounts of information, analyzing it from different vantage points and "proposing and testing models to explain the dynamics of certain behaviors," Stefanescu said, listing such resources as Facebook, LiveJournal, Twitter, and Last.Fm.
The researchers are, among other things, studying the structure of online social networks, looking for properties that may characterize them. These findings may tell scientists how the structure of the network affects members' behavior, as well as suggest patterns of future social networks.
The National Science Foundation has provided $319,917 for the work from the American Recovery and Reinvestment Act of 2009.
This past summer, for example, Stefanescu's students examined certain mathematical and statistical elements of massive online social networks, in particular the direct and indirect relationships among "friends."
"A typical representation of an online social network is a graph in which its members are represented by nodes, and their direct connections by edges in the graph," Stefanescu explained. "In the vernacular, two individuals directly connected are commonly referred to as 'friends.' Two individuals may not be directly connected, but there may exist a path of connections between them. For example, suppose that A, B and C are three nodes—individuals—in the network. Furthermore, suppose that A and B are connected, B and C are connected, but A and C are not connected. Then, while A and B are not 'friends,' there is still a some kind of a relationship between them because A is a 'friend' of a 'friend' of C.