IP Library Granted Patent US 12,609,116
Granted Patent B2
US 12,609,116 · App. 18/330,766 · Granted Apr 21, 2026

Electronic apparatus and method for controlling thereof

Inventors: Hyungtak Choi (Suwon-si, KR); Lohith Ravuru (Suwon-si, KR); Hyeonmok Ko (Suwon-si, KR); Haehun Yang (Suwon-si, KR); Seungchul Lee (Suwon-si, KR)
Assignee: Samsung Electronics Co., Ltd.
G10L15/22G10L2015/223G10L2015/228
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,609,116
App. No.
18/330,766
Granted
Apr 21, 2026
Kind
B2
Abstract

An electronic apparatus is provided. The electronic apparatus includes a communication device, a memory configured to store at least one instruction and one or more vector values corresponding to dialogue history information, and a processor, based on execution of the at least one instruction, configured to extract text from the dialogue content received through the communication device, calculate a vector value of the extracted text by using a predetermined encoding algorithm, and generate response information by using the calculated vector value and the stored one or more vector values.

Claims (70)

1 . An electronic apparatus comprising:

a communication device;

a memory configured to store at least one instruction and one or more vector values corresponding to dialogue history information, wherein the dialogue history information includes at least one of user dialogue history or another user dialogue history; and

a processor, based on execution of the at least one instruction, configured to:

receive dialogue content through the communication device,

extract text from dialogue content,

calculate a vector value of the extracted text by using a predetermined encoding algorithm,

obtain a plurality of candidate vector values each having the predetermined similarity with the calculated vector value among the stored one or more vector values corresponding to dialogue history information,

decode each of the obtained plurality of candidate vector values into the text,

confirm slot information in the decoded text,

confirm a user request and slot information in the extracted text,

generate the response information by using the confirmed user request and the confirmed slot information in the extracted text, and the confirmed slot information in the decoded text, and

transmit the generated response information to an external device through the communication device.

2 . The apparatus as claimed in claim 1 ,

wherein the processor, based on the execution of the at least one instruction, is further configured to:

extract additional information from the received dialogue content,

wherein the vector value is calculated by additionally using the extracted additional information, and

wherein the additional information includes at least one of time information, dialogue frequency, or user emotion information of the dialogue content.

3 . The apparatus as claimed in claim 1 ,

wherein the processor, based on the execution of the at least one instruction, is further configured to:

identify whether additional information is to be used to generate the response information, and

obtain a vector value having predetermined similarity with the calculated vector value among the stored one or more vector values in case of identifying that the additional information is to be used, and

wherein the response information is generated by additionally using the obtained vector value.

4 . The apparatus as claimed in claim 3 , wherein the processor, based on the execution of the at least one instruction, is further configured to:

confirm a user request and slot information by using the extracted text, and

determine whether the response information corresponding to the user request is possible to be obtained using the confirmed slot information.

5 . The apparatus as claimed in claim 1 , wherein the processor, based on the execution of the at least one instruction, is further configured to:

determine a weight for each confirmed slot information by using the dialogue content, and

obtain the vector value having the predetermined similarity with the calculated vector value among the stored one or more vector values by using the determined weight.

6 . The apparatus as claimed in claim 1 , wherein the processor, based on the execution of the at least one instruction, is further configured to:

generate one text phrase by using text corresponding to the vector value having the predetermined similarity and the text extracted from the dialogue content,

calculate a vector of the generated text phrase by using the predetermined encoding algorithm, and

store the calculated vector in the memory.

7 . A method performed by an electronic apparatus, the method comprising:

storing one or more vector values corresponding to dialogue history information, wherein the dialogue history information includes at least one of user dialogue history or another user dialogue history;

receiving dialogue content through a communication device included in the electronic apparatus;

generating response information corresponding to the dialogue content; and

transmitting the generated response information through the communication device,

wherein the generating of the response information includes:

extracting text from the dialogue content;

calculating a vector value of the extracted text by using a predetermined encoding algorithm;

obtaining a plurality of candidate vector values each having the predetermined similarity with the calculated vector value among the stored one or more vector values corresponding to dialogue history information;

decoding each of the obtained plurality of candidate vector values into the text;

confirming slot information in the decoded text;

confirming a user request and slot information in the extracted text; and

generating the response information by using the confirmed user request and the confirmed slot information in the extracted text, and the confirmed slot information in the decoded text.

8 . The method as claimed in claim 7 ,

wherein in the calculating of the vector value, additional information is extracted from the received dialogue content,

wherein the vector value is calculated by additionally using the extracted additional information, and

wherein the additional information includes at least one of time information, dialogue frequency, or user emotion information of the dialogue content.

9 . The method as claimed in claim 7 , wherein the generating of the response information further includes:

determining a weight for each confirmed slot information by using the dialogue content; and

obtaining the vector value having the predetermined similarity with the calculated vector value among the stored one or more vector values by using the determined weight.

10 . The method as claimed in claim 7 , further comprising:

generating one text phrase by using text corresponding to the vector value having the predetermined similarity and the text extracted from the dialogue content;

calculating a vector of the generated text phrase by using the predetermined encoding algorithm; and

storing the calculated vector.

11 . A non-transitory computer-readable recording medium including a program for executing a method for controlling an electronic apparatus, wherein the method includes:

storing one or more vector values corresponding to dialogue history information, wherein the dialogue history information includes at least one of user dialogue history or another user dialogue history;

receiving dialogue content through a communication device included in the electronic apparatus;

generating response information corresponding to the dialogue content; and

transmitting the generated response information through the communication device,

wherein the generating of the response information includes:

extracting text from the dialogue content;

calculating a vector value of the extracted text by using a predetermined encoding algorithm;

obtaining a plurality of candidate vector values each having the predetermined similarity with the calculated vector value among the stored one or more vector values corresponding to dialogue history information;

decoding each of the obtained plurality of candidate vector values into the text;

confirming slot information in the decoded text;

confirming a user request and slot information in the extracted text; and

generating the response information by using the confirmed user request and the confirmed slot information in the extracted text, and the confirmed slot information in the decoded text.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Jun 7, 2023
From: CHOI, HYUNGTAK; RAVURU, LOHITH; KO, HYEONMOK; YANG, HAEHUN; LEE, SEUNGCHUL
To: SAMSUNG ELECTRONICS CO., LTD.
Reel/Frame 063884/0024 →
Priority Claims (2)
KR 10-2022-0056868 · May 9, 2022 · national
KR 10-2022-0110769 · Sep 1, 2022 · national
Continuity (2)
Continuation PCTKR2023005365 · Apr 20, 2023
Related Publication 20230360651A1 · Nov 9, 2023
References Cited (26)
US 9852177B1 · Cheung · 2017 [cited by examiner]
US 10741178B2 · Jo et al. · 2020 [cited by applicant]
US 11194973B1 · Goel · 2021 [cited by examiner]
US 11984126B2 · Choi et al. · 2024 [cited by applicant]
US 20170025125A1 · Alvarez Guevara · 2017 [cited by applicant]
US 20190013017A1 · Kang et al. · 2019 [cited by applicant]
US 20200064444A1 · Regani et al. · 2020 [cited by applicant]
US 20200090651A1 · Tran · 2020 [cited by examiner]
US 20200097496A1 · Alexander · 2020 [cited by examiner]
US 20200097544A1 · Alexander · 2020 [cited by examiner]
US 20200097563A1 · Alexander · 2020 [cited by examiner]
US 20200395008A1 · Cohen et al. · 2020 [cited by applicant]
US 20210034678A1 · Hashimoto et al. · 2021 [cited by applicant]
US 20210142794A1 · Mathias · 2021 [cited by examiner]
US 20220108080A1 · Munavalli · 2022 [cited by examiner]
US 20220310096A1 · Choi et al. · 2022 [cited by applicant]
CN 110188167B · 2021 [cited by applicant]
CN 113505198A · 2021 [cited by applicant]
KR 1020190133931A · 2019 [cited by applicant]
KR 102098003B1 · 2020 [cited by applicant]
KR 1020200105057A · 2020 [cited by applicant]
KR 1020200143991A · 2020 [cited by applicant]
KR 102297849B1 · 2021 [cited by applicant]
KR 1020220020723A · 2022 [cited by applicant]
Stathoulopoulos; Hands-on Tutorials; How to Build a Semantic Search Engine With Transformers and Faiss; Towards Data Science; Nov. 10, 2020. [cited by applicant]
International Search Report with Written Opinion and English translation dated Jul. 24, 2023; International Appln. No. PCT/KR2023/005365. [cited by applicant]