User Attribute Prediction Using App Usage and Sleep Patterns

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing user attribute prediction methods, particularly for demographic information like age and gender, suffer from low accuracy, lack of universality, and robustness, and pose privacy concerns due to reliance on image and voice data.

Innovation Solution

A deep learning-based method that predicts user attributes using APP usage patterns and sleep patterns without requiring image or voice data, employing a deep neural network to extract correlations and using latent variables to learn weights, enhancing prediction accuracy and universality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If image and voice data are collected for user attribute prediction, then prediction accuracy may be improved, but user privacy is compromised and data capture complexity increases

Engineering Contradiction:
Improveprediction accuracyVSAvoiduser privacy concern
Core Design Contradiction:
Measurement precisionVSObject-affected harmful factors

Solution Approach 1:

The patent extracts and removes the harmful elements (image and voice data collection) from the prediction system, replacing them with text-based demographic information and device usage data. This extraction eliminates privacy concerns while maintaining prediction functionality through alternative data sources that do not compromise user confidentiality.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent introduces text-based demographic information and device usage patterns as intermediary data sources that mediate between the need for accurate prediction and user privacy protection. These intermediaries provide sufficient predictive power without requiring direct collection of sensitive biometric data, thus resolving the contradiction between accuracy and privacy.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If multiple data sources including image and voice are used for prediction, then prediction accuracy improves, but device complexity and data processing requirements increase

Engineering Contradiction:
Improveprediction accuracyVSAvoiddata capture and processing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts and removes the complex data capture and processing requirements for image and voice data, retaining only text-based demographic information and simple device usage metrics. This extraction significantly reduces device complexity and data processing requirements while preserving the core prediction functionality.

Inventive Principle:
Principle #2Taking out (Extraction)

3Measurement precision

If extensive user data is collected to improve prediction accuracy, then more comprehensive user profiles can be constructed, but network throughput and data storage requirements increase

Engineering Contradiction:
Improveprediction accuracyVSAvoidnetwork throughput consumption
Core Design Contradiction:
Measurement precisionVSLoss of energy

Solution Approach 1:

The patent extracts and removes the requirement for extensive data collection and transmission, relying instead on compact text-based demographic information and minimal device usage data. This extraction dramatically reduces network throughput consumption and data storage requirements while maintaining sufficient prediction accuracy for constructing user profiles.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP3973418B1Method, apparatus, electronic device and storage medium for predicting user attribute
Publication Date: 2025.10.29 SAMSUNG ELECTRONICS CO LTD
  • EP3973418B1 patent drawingFigure 1~3
  • EP3973418B1 patent drawingFigure 4
  • EP3973418B1 patent drawingFigure 5

AI summary

The embodiment of the present disclosure provides a method, an apparatus, an electronic device, and a storage medium for predicting user attribute. The method comprises acquiring an input of demographic-related information; determining features corresponding to the input of the demographic-related information and a weight corresponding to each feature, where the features comprise a single feature corresponding to each input of the demographic-related information, and/or a cross feature between the at least two single features; predicting user attribute according to each feature and its corresponding weight. Based on the solution provided by the embodiment of the present disclosure, the accuracy on the prediction of the user attribute can be effectively improved.