Linked Activity Spaces: Embedding Social Networks in Urban Space - Computational Approaches for Urban Environments - page 313

shopping centers), public services and institutions (hospitals, post offices, police
stations), transportation centers (airports, train stations), and named villages in the
suburban area. POIs of similar types (e.g., restaurants) in a 50 m radius are grouped
into a single POI to eliminate redundant information.
As mentioned, two activity spaces are considered linked if their corresponding
nodes are connected via the social network. Thus, friendships are embedded in
geographic space through the two activity spaces. Each unique pair of linked
activity spaces (LASs) is compared by the number of common POIs shared by both
(in absolute number and percent of the user's total POIs). In Fig. 13.5 ,twopairs
of linked activity space ellipses share POIs, where the pair demarcated with a solid
blue line shares more POIs. In Fig. 13.6 , an ego ellipse is in focus (in solid pink)
and shares POIs to various extents with each of his or her alters (in dashed lines).
We use a statistical t-test to compare friends' typical number of shared POIs versus
that of a random pair of users. Our hypothesis is that friends share more POI points
than random pairs.
Fig. 13.5 This image shows how linked activity spaces overlap and can be quantified using the
number of POIs found in the intersection of activity spaces. These POIs can be a proxy for
convenience of shared meeting points. More specifically, this figure shows two pairs of linked
activity spaces, in solid blue and dashed red lines . The pair of solid blue ellipses intersects more
commonPOIsthanthetwo dashed red ellipses. However, this set of ellipses has a particular place
of intersection that is west of the most general POI cluster trend