User Attribute Inference from Device Usage Patterns
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems face challenges in accurately determining user attributes, interests, and demographics from personal electronic devices due to data silos and limited access to application usage information, making targeted advertising and consumer product development less effective.
Innovation Solution
A trusted service system that collects and analyzes user and device information from personal electronic devices to infer user interests and demographics by training models based on usage patterns, location data, and application attributes, enabling targeted content delivery and advertising.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If data is collected from multiple sources to improve user attribute determination accuracy, then measurement precision improves, but device complexity increases due to multiple data silos and integration requirements
Solution Approach 1:
The patent segments the complex determination process into distinct modules: a determination engine that receives information from multiple sources (device usage information, environmental information, sensor data), processes them separately through trained models, and integrates results to determine user attributes. This modular segmentation allows accurate multi-source data integration while managing system complexity through organized processing stages.
2Loss of information
If more sensor data and usage information are collected to improve user interest inference, then information completeness improves, but loss of time increases due to extensive data processing requirements
Solution Approach 1:
The patent implements preliminary action by pre-training determination models offline using extensive labeled data before deployment. During runtime, the pre-trained models quickly process incoming sensor data and usage information without requiring extensive real-time computation. This separates the time-consuming training phase from the fast inference phase, ensuring information completeness while minimizing processing time delays.
Solution Approach 2:
The system employs self-service mechanisms where determination models automatically select and weight relevant features from multiple data sources based on their training, without requiring manual feature engineering or real-time data prioritization. The models autonomously process complete information sets efficiently, reducing both information loss and processing time.
3Reliability
If verified label data is used to train determination models, then reliability of user attribute prediction improves, but quantity of substance decreases due to limited availability of verified user information
Solution Approach 1:
The patent merges multiple data sources including device usage information, environmental information, and sensor data to create comprehensive training datasets. By combining these diverse information streams, the system accumulates sufficient training data quantity while maintaining reliability through the use of verified label data where available, supplementing it with inferred data from multiple correlated sources.
Data Source
AI summary
This disclosure relates to systems and methods for determining information associated with a user. In certain embodiments, various attributes, interest, and/or other demographic information related to a user may be determined based on information may be obtained from personal electronic devices associated with the user. User attribute models used to predict user attributes and/or demographics may be generated and trained based on inferred user interests and available label data. In some embodiments, information reflecting which applications and/or types of applications are installed and/or frequently used by a user on their personal electronic devices may be used in connection with determining interests associated with the user. Further embodiments of the systems and methods disclosed herein relate to determining various information associated with a user based on location information indicative of a user's location obtained from one or more personal electronic devices associated with the user.