Voice Query Classification for Privacy Protection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Computer-based services face challenges in detecting and protecting sensitive information within user questions, risking exposure when answering voice-uttered queries, especially when relying on external resources for answers.
Innovation Solution
A system that analyzes voice search queries by converting them to text, matching against known phrases and templates, and evaluating risk scores to determine if the query includes sensitive information, thereby deciding whether to provide it to secondary computer systems for answering, ensuring protection of user privacy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If computer-based services answer user questions using external resources, then the ability to provide accurate answers is improved, but the risk of exposing sensitive information increases
Solution Approach 1:
The system performs preliminary classification of voice search queries to identify sensitive information before submitting them to external resources. This advance detection and categorization prevents sensitive data from being exposed while still allowing non-sensitive queries to be answered accurately using external resources.
Solution Approach 2:
The patent introduces an intermediary classification system that sits between the user's voice query and the external resources. This intermediary layer analyzes and categorizes queries, acting as a mediator that filters out sensitive information before it reaches external systems, thus protecting privacy while maintaining answer accuracy for safe queries.
2Reliability
If computer-based services analyze all voice queries for sensitive information, then privacy protection is improved, but the processing time and system complexity increase
Solution Approach 1:
The classification system is segmented into distinct modules that handle different aspects of query analysis. By dividing the sensitive information detection into separate classification components, the system achieves comprehensive privacy protection while organizing complexity into manageable, independent segments that can be processed efficiently.
Data Source
AI summary
Devices, systems, and methods are provided for classifying voice search queries. A system may receive voice data associated with a voice utterance, the system being associated with a network. The system may determine that the voice data is associated with a question, and may determine an absence of an answer to the question. The system may determine a score associated with the question, the score indicative of a risk of disclosure of sensitive information associated with a person. The system may determine that the score fails to satisfy a threshold, and may send the question to a device, wherein the device is remote from the network. The system may receive data associated with the question.