| 主权项 |
1. A method, comprising:
obtaining geolocations histories of computing devices, the geolocation histories spanning a duration of time exceeding, on average, more than one week, each geolocation history corresponding to a respective individual, the number of geolocation histories exceeding 10,000, the number of geolocation in each geolocation history exceeding, on average, three geolocations, each geolocation history having an identifier of a device or individual and a plurality of timestamped geolocations; assigning different subsets of the geolocation histories to different computing devices in a compute cluster configured for distributed storage and processing of data; querying a geographic information system (GIS) with geolocations in the geolocations histories to obtain identifiers of retail chains associated with at least some of the geolocations, the GIS storing data mapping geographic areas to more than 100 retail chain establishments of more than 10 retail chains; with the respective computing devices operating upon data stored locally by the respective computing device, determining visit graphs for the individuals, each visit graph having an identifier of a respective individual and identifiers of a plurality of retail chains visited by the respective individual as indicated the location history and one or more query responses from the GIS; and determining, for a given retail chain and a given individual, a score indicative of an affinity of the given individual to the given retail chain based on the visit graphs for more than 100 individuals including the given individual. |
