Personalization Identifier for Privacy-Safe Content Delivery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current natural language processing systems face challenges in providing personalized supplemental content to users without compromising user privacy, as skills often require user-identifying data to tailor recommendations, which can lead to privacy concerns and increased latency in processing.
Innovation Solution
The system generates a personalization identifier specific to each user and skill using a one-way hash of the user's profile identifier, skill identifier, and timestamp, allowing for customized content delivery without revealing the user's identity to the skill, and enables parallel processing to minimize latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If user-identifying data is used to tailor supplemental content recommendations, then personalization quality is improved, but user privacy is compromised
Solution Approach 1:
The patent introduces a personalization identifier as an intermediary that mediates between user identity and supplemental content delivery. This identifier is generated from user profile data but does not directly expose the user's permanent identifying information to the skill, thus enabling personalization while protecting privacy.
Solution Approach 2:
The patent creates a copy or representation of user identity through the personalization identifier, which captures enough information to enable personalized content delivery without requiring the skill to access or store the actual user's permanent identifying data.
2Measurement precision
If user-identifying data is used to tailor supplemental content, then recommendation accuracy is improved, but processing latency increases
Solution Approach 1:
The patent performs preliminary actions by generating the personalization identifier during the ASR and NLU processing stages, before the supplemental content selection occurs. This allows the identifier to be ready when needed, avoiding additional processing delays later in the pipeline.
Solution Approach 2:
The patent merges the personalization identifier generation process with the existing ASR and NLU processing workflows, so that the identifier is created as part of the normal speech processing pipeline rather than as a separate, time-consuming step.
3Adaptability or versatility
If permanent user-identifying data is stored and used, then personalized content delivery is enabled, but privacy security is reduced
Solution Approach 1:
The patent extracts only the necessary elements from the user's permanent identifying data to create the personalization identifier, separating the functional requirements for personalization from the sensitive permanent identifying information. This extraction allows personalized content delivery while removing the security risks associated with storing and using complete user identifiers.
Data Source
AI summary
Techniques for generating a personalization identifier that is usable by a skill to customize output of supplemental content to a user, without the skill being able to determine an identity of the user based on the personalization identifier, are described. A personalization identifier may be generated to be specific to a skill, such that different skills receive different personalization identifiers with respect to the same user. The personalization identifier may be generated by performing a one-way hash of a skill identifier, and a user profile identifier and/or a device identifier. User-perceived latency may be reduced by generating the personalization identifier at least partially in parallel to performing ASR processing and/or NLU processing.