IP Library Granted Patent US 12,566,927
Granted Patent B2
US 12,566,927 · App. 18/671,962 · Granted Mar 3, 2026

Dialogue processing apparatus, learning apparatus, dialogue processing method, learning method and program

Inventors: Yasuhito Osugi (Tokyo, JP); Itsumi Saito (Tokyo, JP); Kyosuke Nishida (Tokyo, JP); Hisako Asano (Tokyo, JP); Junji Tomita (Tokyo, JP)
Assignee: NTT, Inc.
G06F40/35G06F40/56G06V30/41
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 12,566,927
App. No.
18/671,962
Granted
Mar 3, 2026
Kind
B2
Abstract

A generation unit that takes a question Q i that is a word sequence representing a current question in a dialogue, a document P used to generate an answer A i to the question Q i , a question history {Q i-1 , . . . , Q i-k } that is a set of word sequences representing k past questions, and an answer history {A i-1 , . . . , A i-k } that is a set of word sequences representing answers to the k questions as inputs, and generates the answer A i by machine reading comprehension in an extractive mode or a generative mode using pre-trained model parameters is provided.

Claims (35)

1 . A dialogue processing apparatus comprising:

a hardware processor; and

a storage storing therein a set of instructions that, when executed by the hardware processor, causes the dialogue processing apparatus to:

receive as inputs:

a word sequence representing a current question in a dialogue;

text to be used to generate an answer to the current question;

a question history that is a set of word sequences representing past questions; and

an answer history that is a set of word sequences representing answers to the past questions; and

generate the answer to the current question, by extractive or generative machine reading comprehension, using pre-trained model parameters.

2 . The dialogue processing apparatus according to claim 1 ,

wherein the hardware processor further causes the dialogue processing apparatus to calculate features related to the question history and features related to the answer history in the text, and

wherein the features calculated with respect to the question history and the features calculated with respect to the answer history in the text are incorporated in features in the text that are used to generate the answer to the current question.

3 . The dialogue processing apparatus according to claim 2 , wherein the hardware processor further causes the dialogue processing apparatus to:

calculate features related to the current question in the text;

combine the features related to the current question, the features related to the question history, and the features related to the answer history in the text; and

generate the answer to the current question, by extractive or generative machine reading comprehension, based on the features in the text that combine the features of and reflect the current question, the question history, and the answer history, and that are used to generate the answer to the current question.

4 . A learning apparatus comprising:

a hardware processor; and

a storage storing therein a set of instructions that, when executed by the hardware processor, causes the learning apparatus to:

receive as inputs:

a word sequence representing a current question in a dialogue;

text to be used to generate an answer to the current question;

a question history that is a set of word sequences representing past questions; and

an answer history that is a set of word sequences representing answers to the past questions;

generate the answer to the current question, by extractive or generative machine reading comprehension, using model parameters; and

update the model parameters, by supervised learning, using the generated answer and a correct answer to the current question.

5 . A dialogue processing method to be executed by a computer, the method comprising:

receiving as inputs:

a word sequence representing a current question in a dialogue;

text to be used to generate an answer to the current question;

a question history that is a set of word sequences representing past questions; and

an answer history that is a set of word sequences representing answers to the past questions; and

generating the answer to the current question, by extractive or generative machine reading comprehension, using pre-trained model parameters.

6 . A non-transitory computer-readable recording medium storing therein a set of instructions that, when executed by a computer, causes the computer to operate as the dialogue processing apparatus of claim 1 .

7 . A non-transitory computer-readable recording medium storing therein a set of instructions that, when executed by a computer, causes the computer to operate as the learning apparatus of claim 4 .

Assignments (2)
CHANGE OF NAME Recorded Oct 22, 2025
From: NIPPON TELEGRAPH AND TELEPHONE CORPORATION
To: NTT, INC.
Reel/Frame 073184/0535 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Feb 6, 2025
From: OSUGI, YASUHITO; SAITO, ITSUMI; NISHIDA, KYOSUKE; ASANO, HISAKO; TOMITA, JUNJI
To: NIPPON TELEGRAPH AND TELEPHONE CORPORATION
Reel/Frame 070135/0201 →
Continuity (2)
Continuation 17613417
Related Publication 20240320440A1 · Sep 26, 2024
References Cited (10)
US 12026472B2 · Osugi · 2024 [cited by examiner]
US 20190278812A1 · Otsuka · 2019 [cited by examiner]
US 20210232948A1 · Otsuka · 2021 [cited by examiner]
US 20220229997A1 · Osugi et al. · 2022 [cited by applicant]
Huang et al. (2019) “FlowQA Grasping Flow in History for Conversational Machine Comprehension”, Published as a conference paper at ICLR 2019, arXiv preprint arXiv: 1810.06683. [cited by applicant]
Devlin et al. (2019) “BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding”, arXiv preprint arXiv: 1810.04805. [cited by applicant]
Vaswani et al. (2017) “Attention is All You Need”, arXiv preprint arXiv: 1706.03762. [cited by applicant]
Seo et al. (2017) “Bidirectional Attention Flow for Machine Comprehension”, in ICLR. [cited by applicant]
See et al. (2017) “Get to the Point: Summarization with Pointer-generator Networks”, in Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (vol. 1: Long Papers), pp. 1073-1083, Jul. … [cited by applicant]
Choi et al. (2018) “QuAC: Question Answering in Context”, in EMNLP. ArXiv: 1808.07036. [cited by applicant]