公开(公告)号：US20200065694A1

公开(公告)日：2020-02-27

申请号：US16107686

申请日：2018-08-21

Applicant: Facebook, Inc.

Inventor： Saurav Mohapatra ,  Vladimir Leonid Bychkovsky ,  Rohit Garg ,  Mostafa Keikha

IPC: G06N7/00 ,  G06F17/30 ,  G06Q30/02 ,  G06Q50/00

Abstract: A method includes receiving location reports indicating locations of mobile devices associated with users of an internet platform, registering a count for each location report, determining, for each location report received from a mobile device, a recent location report received from the mobile device indicating a previous location and registering a transition for each of a paired location report and recent location report, corresponding to a pair of locations. The method includes counting a number of transitions corresponding to a particular pair of locations and determining common transitions by comparing the number of transitions to a threshold value. The method includes comparing a location report received from a user's mobile device with location reports included in common transitions, and predicting, based on the comparison, a likelihood the user will arrive at a particular place within a particular time period or a likelihood that the user was at a particular place within a particular time before the current time.

公开(公告)号：US20200043046A1

公开(公告)日：2020-02-06

申请号：US16054367

申请日：2018-08-03

Applicant: Facebook, Inc.

Inventor： Tsuwei Chen ,  Qizhen Ruan ,  Roy Koonammave Jose ,  Scott J. Bratsman ,  William Bullock ,  Aude Hofleitner ,  Yoav Shapira ,  Mostafa Keikha

IPC: G06Q30/02 ,  G06F17/30 ,  H04W4/021 ,  G06F15/18

Abstract: In one embodiment, a method includes analyzing social graph information associated with users of a social-networking system, developing feature vectors describing elements of social graph information, and applying the feature vectors to determine the relevance of elements of social graph information to the location of special relevance. The method further includes receiving at least one data point from a user's networked device, applying the feature vectors to the at least one data point to determine the relevance of the at least one data point to the location of special relevance, and assigning weight to each data point based on the determined relevance of each data point to the location of special relevance. Finally, the method includes processing the at least one data point according to its assigned weight and forming a prediction, to a particular degree of certainty, indicating the user's location of special relevance.