| Contents |
| In this talked, a model (called Elliptic Model) is presented to estimate the number of social ties between two locations using population data in a similar manner transportation research does with trips. To overcome the asymmetry of transportation models, the new model considers that the number of relationships between two locations is reversely proportional to the population in the ellipse whose focuses are in these two locations. The Elliptic Model is evaluated by considering the anonymous communications patterns of 25 million users from 3 different countries, where a location has been assigned to each user based on her most used phone tower or billing zip code. With this information, spatial social networks are built at three levels of resolution: tower, city and region for each of the three countries. The Elliptic Model reaches similar performance when predicting communication fluxes as transportation models do when predicting trips. This shows that human relationships are at least as influenced by geography as human mobility is |
|