User Group Interest Modeling for Privacy-Safe Content Distribution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for selecting and delivering personalized digital components face challenges in maintaining user privacy due to the unavailability of user-specific data, such as third-party cookies, while ensuring efficient and accurate content distribution.
Innovation Solution
A machine learning approach that assigns users to user groups based on their digital content access patterns, generating training datasets from group and user-specific data to predict user interests without using individual identifiers, employing transfer learning techniques to select and distribute personalized digital components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If user-specific data is collected and transmitted for personalized content delivery, then personalization accuracy is improved, but user privacy is compromised and data transmission requirements increase
Solution Approach 1:
The patent segments users into anonymous user groups based on shared characteristics and interests rather than treating individual users separately. This segmentation allows the system to deliver personalized content to groups while maintaining individual user privacy, as the personalization is based on group-level patterns rather than individual data transmission.
Solution Approach 2:
The patent introduces an intermediary machine learning model that processes user group data and generates personalized content recommendations without requiring direct transmission of individual user-specific data. The model acts as a mediator between user group characteristics and content delivery, enabling personalization while preserving privacy.
2Productivity
If traditional content selection methods are used without machine learning models, then computational resources are conserved, but content delivery time increases and personalization capability decreases
Solution Approach 1:
The patent performs preliminary training of machine learning models using user group data before actual content delivery. This preliminary action creates pre-computed models that can quickly generate personalized content recommendations during runtime, reducing content delivery time while the computational cost is amortized over the training phase rather than real-time operations.
3Loss of information
If third-party cookies are blocked to protect user privacy, then user privacy is improved, but the capability to deliver personalized content deteriorates
Solution Approach 1:
The patent inverts the traditional approach by not tracking individual users through cookies, but rather creating anonymous user groups based on shared characteristics. Instead of following the individual user across the web, the system inverts to group users who share similar interests and behaviors, enabling personalization without individual tracking.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training and using machine learning models. In some aspects, a method includes identifying a first set of data for users of multiple user groups. For each user, a first party user identifier is obtained that identifies the individual user to a first party content provider. A second set of data describing activity of the user with respect to content of the first party content provider is identified. For each user, a contextual analysis of the first set and the second set of data is performed to generate one or more labels indicating user interest. A training dataset is generated based on the first set and the second set of data and a label. The training dataset is then used to train one or more machine learning models to predict user interest.