A system and method for pacing information campaigns based on predicted and observed position events

摘要

Problem to be solved: to provide a system and method for pacing an information campaign using predicted and observed position events.Location prediction system 100 includes a machine learning module for training a position prediction model for an information campaign, a front end server for receiving and processing an information campaign, and a prediction unit 130.During an information campaign, the prediction unit uses a location prediction model to predict the conversion probability for any particular mobile device associated with the passed information campaigns received at any given time unit.The conversion probability corresponds to the probability that a particular mobile device has at least one positional event in one or more POI in a particular time frame.The front-end server further determines each target conversion number to be achieved by the information campaign in each time unit, and determines a response to a particular information request based on the conversion probability and each target conversion number.Diagram