A New Model of Social Networks
Recently, Heather Mattie and colleagues at Harvard University have released a new study called “The Social Bow Tie” which analyzes the distributions and strength of relationships in social networks. The researchers studied two social networks: the network of 70,000 people in 75 rural villages in India, and 50,000 mobile phone users from a European country. In both networks, they analyzed and constructed ties of varying strength between people based on factors such as regular interactions (more interactions = stronger tie). For each pair of people in the network, they mapped out the network of mutual friends as well as the friends that they did not have in common, resulting in a bow-tie formation, as seen below.
Nodes i and j on the graph are the two people analyzed, and each edge connects nodes to their friends. The blue circle highlights the network of friends that they had in common, and the red area are friends that they do not have in common. Machine learning and regression methods were able to show that the amount of mutual-friends in the bow tie structure was the most predictive of tie strength.
This study encapsulates many ideas about ties between individuals, with one being from Mark Granovetter’s “The Strength of Weak Ties” which discusses Strong Triadic Closure, as studied in class. This study, however, shows that measuring the strength of links between individuals can be done by analyzing the structure of their social network. In other words, individuals are more strongly linked with people that they have mutual friends with: the center of the Bow Tie. In the same way as the Strong Triadic Closure property, more common friends implies a stronger tie.
“How Close Are You Really?” https://www.technologyreview.com/s/609146/how-close-are-you-really/
“The Social Bow Tie” https://arxiv.org/abs/1710.04177#