On-device content personalization preserving user privacy

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing content delivery systems risk exposing user privacy by distributing personal preference information outside of the user's device, compromising user privacy.

Innovation Solution

A computing system that delivers personalized content by filtering, deduplicating, and clustering content items relevant to an investment identifier on a content server, and then selecting and presenting personalized content items on the user device without sharing user preference data externally.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If user preference information is distributed to content servers for personalized content delivery, then content personalization quality is improved, but user privacy is compromised

Engineering Contradiction:

Improvecontent personalization qualityVSAvoiduser privacy exposure

Core Design Contradiction:

Adaptability or versatilityVSObject-affected harmful factors

Solution Approach 1:

Instead of sending user preference data to the content server for processing, the patent inverts the approach by sending only content metadata to the user device, where the user device performs the personalization evaluation locally using stored preference information. This reversal of data flow direction eliminates privacy exposure while maintaining personalization quality.

Inventive Principle:

Principle #13The other way round (Inversion)

Solution Approach 2:

The patent extracts and removes sensitive user preference information from the data transmission process. Only non-sensitive content metadata (titles, descriptions, investment identifiers) is sent to the user device, while the sensitive preference data remains exclusively on the user device, enabling personalization without exposing user information.

Inventive Principle:

Principle #2Taking out (Extraction)

2Productivity

If content filtering and clustering is performed on the content server, then processing efficiency is improved, but user device computational load increases

Engineering Contradiction:

Improvecontent processing efficiencyVSAvoiduser device computational energy

Core Design Contradiction:

ProductivityVSUse of energy by moving object

Solution Approach 1:

The patent segments the content delivery process into two distinct phases: (1) Server-side content preparation including filtering, deduplication, and clustering based on investment identifiers; (2) Client-side personalization evaluation using stored preferences. This segmentation allows computationally intensive operations to be performed on the server while minimizing user device processing requirements.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The content server performs preliminary actions by pre-processing content items before transmission - filtering out irrelevant content, removing duplicates, and organizing into clusters. This preliminary processing reduces the volume and complexity of data that needs to be evaluated on the user device, thereby reducing computational energy consumption.

Inventive Principle:

Principle #10Preliminary action

3Measurement precision

If all content items are transmitted to user device for local evaluation, then content selection accuracy is improved, but network bandwidth consumption increases

Engineering Contradiction:

Improvecontent selection accuracyVSAvoidnetwork bandwidth energy

Core Design Contradiction:

Measurement precisionVSLoss of energy

Solution Approach 1:

The patent extracts and transmits only the essential content metadata (titles, descriptions, investment identifiers, cluster identifiers) to the user device, omitting unnecessary data. This selective extraction maintains the information needed for accurate local evaluation while minimizing network bandwidth consumption.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The transmitted content metadata serves multiple functions: it enables local personalization evaluation, allows for duplicate detection through cluster identifiers, and provides sufficient information for accurate content selection without requiring transmission of complete content items or additional processing data.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS20250148029A1On device personalization of content to protect user privacy

Publication Date: 2025.05.08 APPLE INC

AI summary

In some implementations, a computing system can deliver personalized content while preserving user privacy. For example, the computing system can include a content server that filters, deduplicates, and generates clusters of content items (e.g., articles, news stories, etc.) that are relevant to an investment identifier received in a request from a client device. The content server can send the clusters of content items to the requesting client. Upon receiving the clusters of content items, requesting client device can evaluate the content items based on user preferences stored on the user device and select a representative content item from one or more content item clusters. The selected content item can then be presented on a display of the user device. Thus, personalization of content item selection and presentation can be performed without distributing user preference data outside of the user device thereby preserving user privacy.