Local Content Personalization Engine for Privacy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional content identification systems using collaborative filtering and matrix factorization expose user interactions and personalized data to servers, making them vulnerable to data breaches and unwanted use.
Innovation Solution
A method where a local content personalization engine on a user device determines affinity values and classification codes for content items, creates an affinity vector, and transmits it to a server, while keeping interaction data local and transforming it into classification codes for privacy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional collaborative filtering and matrix factorization are used for content identification, then personalized content recommendations can be generated, but user interaction data is exposed to servers making it vulnerable to data breaches
Solution Approach 1:
The patent extracts only the essential affinity information from user interaction data and transmits it to the server, while keeping the raw interaction data local. This selective extraction allows personalized recommendations to be generated without exposing sensitive user interaction details to external servers, thereby reducing data breach vulnerability while maintaining recommendation quality.
Solution Approach 2:
The patent introduces an intermediary processing layer that transforms raw user interaction data into affinity values before transmission. This intermediary transformation acts as a buffer that protects the original sensitive data while still enabling the server to perform collaborative filtering and generate personalized content recommendations.
2Measurement precision
If raw interaction data is transmitted to the server for collaborative filtering, then accurate content recommendations can be generated, but the amount of personal data shared increases
Solution Approach 1:
The system extracts only the necessary affinity metrics from complete interaction data, transmitting minimal information to the server. This extraction process maintains recommendation accuracy by preserving the essential preference signals while discarding extraneous personal details that would increase data exposure risks.
Solution Approach 2:
The patent transforms interaction data from its original detailed form into aggregated affinity values with different parameter characteristics. This parameter transformation reduces the information content to only what is necessary for recommendation accuracy, thereby minimizing personal data exposure while maintaining the precision needed for effective collaborative filtering.
Data Source
AI summary
Disclosed examples can relate to obtaining identifications of content (e.g., content recommendations) while keeping at least some interaction data locally private. For a given user and device, content items for which the user may have an affinity can be predicted based on the interactions of the user with other content items. Respective interaction data for respective content items can stay local to the user device by transforming the respective content items into content codes (e.g., determined based on a codebook generated by clustering perceptual values). The affinity for content codes can be transmitted to the server for use in determining identifications of content items to provide to the device.