Processing

Please wait...

Settings

Settings

Goto Application

1. WO2020111314 - CONCEPTUAL GRAPH-BASED QUERY-RESPONSE APPARATUS AND METHOD

Publication Number WO/2020/111314
Publication Date 04.06.2020
International Application No. PCT/KR2018/014860
International Filing Date 28.11.2018
IPC
G06F 16/00 2019.01
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
FELECTRIC DIGITAL DATA PROCESSING
16Information retrieval; Database structures therefor; File system structures therefor
G06F 17/27 2006.01
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
FELECTRIC DIGITAL DATA PROCESSING
17Digital computing or data processing equipment or methods, specially adapted for specific functions
20Handling natural language data
27Automatic analysis, e.g. parsing, orthograph correction
CPC
G06F 16/00
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
FELECTRIC DIGITAL DATA PROCESSING
16Information retrieval; Database structures therefor; File system structures therefor
G06F 40/20
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
FELECTRIC DIGITAL DATA PROCESSING
40Handling natural language data
20Natural language analysis
Applicants
  • 한국과학기술원 KOREA ADVANCED INSTITUTE OF SCIENCE AND TECHNOLOGY [KR]/[KR]
Inventors
  • 맹성현 MYAENG, Sung Hyon
  • 김경민 KIM, Kyung-Min
Agents
  • 유미특허법인 YOU ME PATENT AND LAW FIRM
Priority Data
10-2018-014801327.11.2018KR
Publication Language Korean (KO)
Filing Language Korean (KO)
Designated States
Title
(EN) CONCEPTUAL GRAPH-BASED QUERY-RESPONSE APPARATUS AND METHOD
(FR) APPAREIL ET PROCÉDÉ D'INTERROGATION-RÉPONSE BASÉS SUR UN GRAPHE CONCEPTUEL
(KO) 개념 그래프 기반 질의응답 장치 및 방법
Abstract
(EN)
A query-response apparatus comprises: a weight application unit for extracting a corpus context word from a corpus, applying a weight to the corpus context word according to a weight rule, extracting a query context word from an input query, and applying a weight to the query context word according to the weight rule; a context similarity determination unit for generating a plurality of corpus sentence embedding vectors for the weight-applied corpus and a query sentence embedding vector for the weight-applied input query, and determining a target sentence embedding vector from among the plurality of corpus sentence embedding vectors on the basis of vector similarities between the plurality of corpus sentence embedding vectors and the query sentence embedding vector; a candidate response determination unit for determining a target conceptual graph corresponding to the target sentence embedding vector, calculating a graph matching score for each of one or more concepts included in the target conceptual graph, and determining one or more candidate concepts corresponding to the input query; and a response determination unit for determining a preset number of concepts from among the candidate concepts as a response to the input query.
(FR)
L’invention concerne un appareil d’interrogation-réponse comprenant : une unité d'application de poids permettant d’extraire un mot contextuel de corpus d'un corpus, d’appliquer un poids au mot contextuel du corpus selon une règle de pondération, d’extraire un mot contextuel d’interrogation d'une interrogation d'entrée, et d’appliquer un poids au mot contextuel de l’interrogation selon la règle de pondération ; une unité de détermination de similarité de contexte permettant de générer une pluralité de vecteurs d’intégration de phrase de corpus pour le corpus pondéré, ainsi qu’un vecteur d’intégration de phrase d'interrogation pour l’interrogation d'entrée pondérée, et de déterminer un vecteur d’intégration de phrase cible parmi la pluralité de vecteurs d’intégration de phrase de corpus d’après les similarités de vecteurs entre la pluralité de vecteurs d’intégration de phrase de corpus et le vecteur d’intégration de phrase d'interrogation ; une unité de détermination de réponse candidate permettant de déterminer un graphe conceptuel cible correspondant au vecteur d’intégration de phrase cible, de calculer un score de correspondance de graphe pour un concept ou chaque concept d’une pluralité de concepts inclus dans le graphe conceptuel cible, et de déterminer un ou plusieurs concepts candidats correspondant à l’interrogation d'entrée ; et une unité de détermination de réponse permettant de déterminer un nombre prédéfini de concepts parmi les concepts candidats en réponse à l’interrogation d'entrée.
(KO)
질의응답 장치로서, 말뭉치에서 말뭉치 맥락 단어를 추출하고, 가중치 규칙에 따라 상기 말뭉치 맥락 단어에 가중치를 적용하고, 입력 질의에서 질의 맥락 단어를 추출하고, 상기 가중치 규칙에 따라 상기 질의 맥락 단어에 가중치를 적용하는 가중치 적용부, 가중치가 적용된 말뭉치에 대한 복수의 말뭉치 문장 임베딩 벡터들 및 가중치가 적용된 입력 질의에 대한 질의 문장 임베딩 벡터를 생성하고, 상기 복수의 말뭉치 문장 임베딩 벡터들과 상기 질의 문장 임베딩 벡터 사이의 벡터 유사도를 기준으로 상기 복수의 말뭉치 문장 임베딩 벡터들 중에서 대상 문장 임베딩 벡터를 결정하는 맥락 유사도 결정부, 상기 대상 문장 임베딩 벡터에 대응하는 대상 개념 그래프를 결정하고, 상기 대상 개념 그래프에 포함된 적어도 하나 이상의 개념들 각각에 대한 그래프 매칭 점수를 계산하여, 상기 입력 질의에 대응하는 적어도 하나 이상의 후보 개념들을 결정하는 후보 응답 결정부, 그리고 상기 후보 개념들 중에서 미리 설정된 수만큼의 개념을 상기 입력 질의에 대한 응답으로 결정하는 응답 결정부를 포함한다.
Also published as
Latest bibliographic data on file with the International Bureau