Sentiment Score Extraction for Privacy-Preserving Server Communication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for interactive communication between a server and a user device require significant computing power and may compromise user privacy due to the transmission of feedback data containing personal information, necessitating additional safety measures.
Innovation Solution
Determine a sentiment score on the user device based on user reactions to notifications, which is then transmitted to the server, reducing server computing burden and preserving user privacy by anonymizing the data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If feedback data containing personal information is transmitted to the server for improving the reinforcement learning model, then the accuracy and appropriateness of notifications is improved, but user privacy is compromised and additional safety measures are required
Solution Approach 1:
The patent extracts only the essential sentiment information from the user's reaction data while leaving the personal identifiers and sensitive information on the user device. Only the anonymized sentiment score is transmitted to the server, separating the useful feedback signal from the privacy-sensitive data.
Solution Approach 2:
The sentiment score acts as an intermediary representation that bridges the user's personal reaction data and the server's learning model. It transforms detailed personal information into an abstracted form that preserves the feedback value while eliminating privacy risks.
2Measurement precision
If the server processes all reaction data to improve the reinforcement learning model, then the model accuracy is improved, but the computing power requirement and server burden increase significantly
Solution Approach 1:
The patent segments the data processing task between the user device and the server. The user device performs local processing to extract the sentiment score from reaction data, while the server only receives and processes the compact sentiment score. This division reduces the computational burden on the server while maintaining model improvement capability.
Solution Approach 2:
The user device performs preliminary processing of the reaction data to extract the sentiment score before transmission to the server. This advance processing reduces the amount of data that needs to be handled by the server, optimizing the overall system efficiency.
3Loss of information
If detailed reaction data is transmitted to the server, then the feedback quality for improving the model is enhanced, but the data transmission volume and processing overhead increase
Solution Approach 1:
The patent extracts only the essential sentiment information from the user's reaction data while leaving the personal identifiers and sensitive information on the user device. Only the anonymized sentiment score is transmitted to the server, separating the useful feedback signal from the privacy-sensitive data.
Data Source
Figure 1~2
Figure 3~4
Figure 5~7
AI summary
Methods and apparatuses for interactive communication between a user device and a server are described. Therein, one or more notifications are provided to a user on the user device, and reaction data indicative of a reaction of the user to the one or more notifications is acquired. Further, based at least in part on the acquired reaction data, a plurality of sentiment scores is determined and, based on comparing at least two of said sentiment scores, a final sentiment score is determined.