Linked Activity Spaces: Embedding Social Networks in Urban Space

We examine the likelihood that a pair of sustained telephone contacts (e.g. friends, family, professional contacts) use the city similarly. Using call data records from an undisclosed city in China, we define a proxy for the daily activity spaces of each individual subscriber by interpolating the points of geo-­‐located cell towers he or she uses most frequently. We calculate the overlap of pairs of linked activity spaces, (e.g. the pairs of space between two established telephone contacts) and find that friends and second degree friends are more likely to overlap than random pairs. We find that higher degree users and users with many network triangles (connected groups of three nodes) tend to congregate in the central business district. We also find that the downtown area hosts many heterogeneous modular communities of social groups (derived from telephone calls), but that two distinct neighborhoods contain distinct social clusters and act like a boundary around these friendships. We connect our findings with the role of social capital in urban planning.