公开(公告)号：US12159118B2

公开(公告)日：2024-12-03

申请号：US18544209

申请日：2023-12-18

Applicant: 42Maru Inc. ,  Korea Advanced Institute of Science and Technology

Inventor： Dong Hwan Kim ,  Sung Ju Hwang ,  Seanie Lee ,  Dong Bok Lee ,  Woo Tae Jeong ,  Han Su Kim ,  You Kyung Kwon ,  Hyun Ok Kim

IPC: G06F40/35 ,  G06N3/045 ,  G06N3/08 ,  G06N5/04

Abstract: The present invention relates to a context-based QA generation architecture, and an object of the present invention is to generate diverse QA pairs from a single context. To achieve the object, the present invention includes a latent variable generating network including at least one encoder and an artificial neural network (Multi-Layer Perceptron: MLP) and configured to train the artificial neural network using a first context, a first question, and a first answer, and generate a second question latent variable and a second answer latent variable by applying the trained artificial neural network to a second context, an answer generating network configured to generate a second answer by decoding the second answer latent variable, and a question generating network configured to generate a second question based on a second context and the second answer.

公开(公告)号：US11960838B2

公开(公告)日：2024-04-16

申请号：US17120075

申请日：2020-12-11

Applicant: 42Maru Inc.

Inventor： Dong Hwan Kim ,  Han Su Kim ,  Woo Tae Jeong ,  Ki Bong Sung ,  Hyeon Dey Kim

IPC: G06F40/279 ,  G06F40/35 ,  G06N3/08

CPC classification number: G06F40/279 ,  G06F40/35 ,  G06N3/08

Abstract: The present invention relates to a method for reinforcing a multiple-choice QA model based on adversarial learning techniques, wherein incorrect answers are further generated based on a data set used in the process of training the multiple-choice QA model to enrich data which are learnable by the multiple-choice QA model. To achieve this object, the method includes step A of an incorrect answer generation model encoding a text based on natural language text and a question, generating a second incorrect answer based on the text and the question, and transmitting the second incorrect answer to an incorrect answer test model, step B of the incorrect answer test model encoding the text, the question, a first correct answer corresponding to the text and the question, a first incorrect answer and the second incorrect answer, and selecting a second correct answer based on results of the encoding, step C of the incorrect answer test model generating a feedback by determining whether the first correct answer is identical to the second correct answer, and step D of the incorrect answer generation model and the incorrect answer test model performing self-learning based on the feedback.

公开(公告)号：US12130851B2

公开(公告)日：2024-10-29

申请号：US18209703

申请日：2023-06-14

Applicant: 42Maru Inc.

Inventor： Dong Hwan Kim ,  Han Su Kim ,  Woo Tae Jeong ,  Seung Hyeon Lee ,  Chang Hyeon Lim

IPC: G06F16/34 ,  G06F16/33 ,  G06F16/901 ,  G06F40/279 ,  G06F40/30 ,  G06F40/40

CPC classification number: G06F16/345 ,  G06F16/3347 ,  G06F16/9024 ,  G06F40/279 ,  G06F40/30 ,  G06F40/40

Abstract: The invention relates to a method and a system for improving performance of text summarization and has an object of improving performance of a technique for generating a summary from a given paragraph. According to the invention to achieve the object, a method for improving performance of text summarization includes: an a step of generating an embedding vector by vectorizing a natural language-based context; a b step of generating a graph using the embedding vector and calculating a first likelihood of each of at least one node included in the graph; a c step of generating a second likelihood by assigning a weight to the first likelihood according to a result of comparing at least one node included in the graph with the context; and a d step of calculating a third likelihood for all candidate paths present in the graph based on the second likelihood, selecting a path having a highest third likelihood, and generating a summary based on the path.

公开(公告)号：US11727041B2

公开(公告)日：2023-08-15

申请号：US17125991

申请日：2020-12-17

Applicant: 42Maru Inc.

Inventor： Dong Hwan Kim ,  Han Su Kim ,  Woo Tae Jeong ,  Seung Hyeon Lee ,  Chang Hyeon Lim

IPC: G06F40/279 ,  G06F40/40 ,  G06F40/30 ,  G06F16/34 ,  G06F16/33 ,  G06F16/901

CPC classification number: G06F16/345 ,  G06F16/3347 ,  G06F16/9024 ,  G06F40/279 ,  G06F40/30 ,  G06F40/40

Abstract: The invention relates to a method and a system for improving performance of text summarization and has an object of improving performance of a technique for generating a summary from a given paragraph. According to the invention to achieve the object, a method for improving performance of text summarization includes: an a step of generating an embedding vector by vectorizing a natural language-based context; a b step of generating a graph by using the embedding vector; a c step of assigning a weight depending on whether or not a keyword corresponding to at least one node included in the graph is present in the context; and a d step of selecting a path having a highest likelihood in the graph and generating a summary based on the path.

公开(公告)号：US11886233B2

公开(公告)日：2024-01-30

申请号：US17096767

申请日：2020-11-12

Applicant: 42Maru Inc. ,  Korea Advanced Institute of Science and Technology

Inventor： Dong Hwan Kim ,  Sung Ju Hwang ,  Seanie Lee ,  Dong Bok Lee ,  Woo Tae Jeong ,  Han Su Kim ,  You Kyung Kwon ,  Hyun Ok Kim

IPC: G06F40/35 ,  G06N3/08 ,  G06N5/04 ,  G06N3/045

CPC classification number: G06F40/35 ,  G06N3/045 ,  G06N3/08 ,  G06N5/04

Abstract: The present invention relates to a context-based QA generation architecture, and an object of the present invention is to generate diverse QA pairs from a single context. To achieve the object, the present invention includes a latent variable generating network including at least one encoder and an artificial neural network (Multi-Layer Perceptron: MLP) and configured to train the artificial neural network using a first context, a first question, and a first answer, and generate a second question latent variable and a second answer latent variable by applying the trained artificial neural network to a second context, an answer generating network configured to generate a second answer by decoding the second answer latent variable, and a question generating network configured to generate a second question based on a second context and the second answer.