Inferring Relationships from Trajectory Data

Devices like smart phones and GPS navigators are very popular nowadays. These equipments can save the location of an object with an associated time, generating a new kind of data, called trajectories of moving objects. With these data it is possible to discover several interesting patterns, among which is the interaction between individuals, allowing to infer their relationship. This work addresses the discovery of relationship degree between moving objects based on their encounters. To calculate the relationship degree we propose different measures based on frequency, duration, and area of the encounters. These measures were evaluated in experiments with a running example and real trajectory data, and show that the method correctly infers relationships.