Location Pair Prediction for Targeted Advertising
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Businesses face difficulties in determining previous locations visited by customers and accurately predicting future locations to provide targeted services, such as advertising and anticipating demand.
Innovation Solution
An online system correlates pairs of locations visited by users, generating location pairs based on sequential lists of locations and predicting high-probability future locations, which are then used for targeted advertising and business decision-making.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If businesses collect and analyze location data from customer devices, then prediction accuracy of future locations improves, but customer privacy concerns and data security risks worsen
Solution Approach 1:
The patent extracts only the essential location pattern information needed for predictions while removing or anonymizing personally identifiable details. The system processes location data to extract movement patterns and destination preferences without retaining or exposing raw location histories, thus achieving accurate predictions while mitigating privacy concerns.
Solution Approach 2:
The patent introduces an intermediary processing layer that acts as a buffer between raw location data and prediction outputs. This intermediary system aggregates and anonymizes location information from multiple users, transforming personal location data into generalized movement patterns that can be used for predictions without exposing individual privacy.
2Reliability
If the system processes location information from multiple users to generate location pairs, then prediction reliability improves, but computational complexity and data processing requirements worsen
Solution Approach 1:
The patent segments the complex task of location prediction into distinct processing stages: collecting location data, extracting location chains, generating location pairs, and making predictions. By dividing the process into manageable segments, the system reduces computational complexity at each stage while maintaining overall prediction reliability through the cumulative effect of each processing step.
Solution Approach 2:
The patent applies partial action by focusing computational resources on processing only the most relevant location data - specifically, recent location chains and frequently visited location pairs. Rather than processing all historical location data equally, the system prioritizes recent and recurrent patterns, reducing computational burden while maintaining prediction reliability.
Data Source
AI summary
An online system receives location information from a plurality of user devices used by users of the online system. The location information identifies a plurality of different locations at which each of the user devices was located. From the location information, a plurality of chains of locations visited by each of a plurality of users are extracted. The online system generates one or more location pairs based on the chain of locations, where each location pair includes a first location and a second location to which there is a high probability a user will travel if the user is located at the first location. The location pairs are used for a variety of applications, such as for advertising to users based on locations and for providing insights into the movements of users.