Consent-Governed Inference Sharing for Personal Assistant Personalization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems fail to effectively share user information between applications, such as web and mobile applications, and smart speaker devices, limiting personalized experiences and efficient task completion.
Innovation Solution
A personalization framework that utilizes a virtual personal assistant to collect user inferences, store them securely, and share them with modules upon user consent, enabling personalized experiences across various applications and devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If user information is shared between applications, then personalized experiences are improved, but user privacy and security control are worsened
Solution Approach 1:
The patent introduces an inference store as an intermediary component that decouples the virtual personal assistant from modules. The inference store stores user inferences and manages access permissions, acting as a secure mediator that allows modules to access personalized data without direct connections to the user's private information, thus maintaining both personalization and privacy control
Solution Approach 2:
The system segments user information access by separating the inference store into multiple isolated inference groups, each with its own permission settings. This segmentation allows different modules to access only the specific inference groups they are authorized to access, enabling personalized experiences for each module while maintaining granular privacy control over different types of user information
2Productivity
If user inferences are stored centrally, then information sharing efficiency is improved, but system complexity is worsened
Solution Approach 1:
The inference store serves multiple functions within a single component: it stores user inferences, manages permission settings for different modules, provides centralized access control, and enables information sharing across diverse applications. This multi-functionality improves information sharing efficiency while avoiding the need for separate complex systems for each function
3Productivity
If modules access user preferences, then user interaction efficiency is improved, but security risks are worsened
Solution Approach 1:
The system performs preliminary security verification by checking permission settings before allowing any module to access user inferences. The inference store verifies whether a module has permission to access specific inference groups before providing data, preventing unauthorized access while enabling efficient legitimate access to user preferences
Data Source
AI summary
A personalization framework is provided. In example embodiments, a machine, including processing circuitry and memory, determines, based on interaction of a user with a virtual personal assistant at one or more client devices associated with an account of the user, a plurality of inferences about the user. The machine stores the plurality of inferences in an inference store associated with the virtual personal assistant. The machine stores user consent data representing whether the user provided consent for a module to access at least a portion of the inferences in the inference store. The machine receives, from the module, a request for a specified inference from the inference store. The machine verifies the user consent data associated with the specified inference and the module. The machine provides the specified inference to the module in response to verifying the user consent data.