On device-based system and method for suppressing leakage of personal information and for providing personalized response
AssigneeSeoul National University R&Db Foundation
InventorsSung Roh Yoon, Sang Won Yu
Life of this patent
Estimating the term…
Abstract
According to an embodiment, an on device-based system for suppressing the leakage of personal information and for providing personalized response includes: a plurality of user devices configured to detect Personal Identifiable Information (PII) from a user query and transmit a user-neutral query which is obtained by converting the PII into neutral information; and a management server configured to receive the user-neutral query and train a management language model that generates a common response pattern for each neutral query pattern.
Claims10
An on-device system for suppressing a leakage of personal information and for providing a personalized response, comprising:
- a plurality of user devices configured to execute a local Personal Identifiable Information (PII) detection model to detect a PII from a user query, locally convert the PII into neutral information by replacing detected PII entities with neutral tokens based on Named Entity Recognition (NER), and transmit a user-neutral query comprising a linguistic distribution of the neutral tokens;
- and
- a management server configured to receive the user-neutral query and train a management language model using federated continual learning that generates a common response pattern for each neutral query pattern, wherein the management server updates the management language model by identifying locations of the neutral tokens within a sentence structure of the user-neutral query to derive common response patterns without ever receiving or storing raw PII from the plurality of user devices.
An on device-based user device for suppressing a leakage of personal information and for providing a personalized response, comprising:
- a communication module;
- a memory to store a personalized response program;
- and
- a processor to execute the personalized response program to:
- detect a PII from a user query;
- locally convert the PII into neutral information by replacing detected PII entities with neutral tokens based on Named Entity Recognition (NER);
- and
- transmit a user-neutral query comprising a linguistic distribution of the neutral tokens to a management server,
- wherein the management server is configured to receive the user-neutral query and train a management language model using federated continual learning that generates a common response pattern for each neutral query pattern, and
- wherein the management server updates the management language model by identifying locations of the neutral tokens within a sentence structure of the user-neutral query to derive common response patterns without ever receiving or storing raw PII from a plurality of user devices.
A personalized response method that is performed by an on device-based user device for suppressing a leakage of personal information and for providing a personalized response, comprising:
- (a) receiving, by a management server, a user-neutral query, which is obtained by converting PII contained in a user query into neutral information from a plurality of user devices, wherein the plurality of user devices execute a local Personal Identifiable Information (PII) detection model to detect the PII from the user query, locally convert the PII into the neutral information by replacing detected PII entities with neutral tokens based on Named Entity Recognition (NER), and transmit a user-neutral query comprising a linguistic distribution of the neutral tokens to the management server;
- and
- (b) training, by the management server, a management language model using federated continual learning that generates a common response pattern for each neutral query pattern, wherein the management server updates the management language model by identifying locations of the neutral tokens within a sentence structure of the user-neutral query to derive common response patterns without ever receiving or storing raw PII from the plurality of user devices.
Dependencies are inferred from the claim text.