Traitement en cours

Veuillez attendre...

Paramétrages

Paramétrages

Aller à Demande

1. CN108763495 - Man-machine dialog method and system, electronic equipment and storage medium

Office Chine
Numéro de la demande 201810536651.0
Date de la demande 30.05.2018
Numéro de publication 108763495
Date de publication 06.11.2018
Numéro de délivrance 108763495
Date de délivrance 20.09.2019
Type de publication B
CIB
G06F 16/90
GPHYSIQUE
06CALCUL; COMPTAGE
FTRAITEMENT ÉLECTRIQUE DE DONNÉES NUMÉRIQUES
16Recherche d’informations; Structures de bases de données à cet effet; Structures de systèmes de fichiers à cet effet
90Détails des fonctions des bases de données indépendantes des types de données cherchés
Déposants AI SPEECH LTD.
苏州思必驰信息科技有限公司
Inventeurs CHEN LU
陈露
CHU MIN
初敏
YANG CHAO
杨超
GE FUJIANG
葛付江
GUO TAOTAO
郭涛涛
Mandataires 北京商专永信知识产权代理事务所(普通合伙) 11400
北京商专永信知识产权代理事务所(普通合伙) 11400
Titre
(EN) Man-machine dialog method and system, electronic equipment and storage medium
(ZH) 人机对话方法、系统、电子设备及存储介质
Abrégé
(EN)
The invention discloses a man-machine dialog method applied to a man-machine dialog system. The method comprises the steps that the current state of the man-machine dialog system is used as input of afirst neural network to determine an active dialog mode of the man-machine dialog system with a user; to-be-recommended topics are determined according to the active dialog mode; the current state and feature vectors of the to-be-recommended topics are used as input of a second neural network to determine recommendation probabilities of the to-be-recommended topics; and to-be-recommended knowledge points are selected from the to-be-recommended topics according to recommendation probability values and displayed to the user. According to the embodiment, the active dialog mode with the user is determined according to the current state of the man-machine dialog system, therefore, it can be guaranteed that the form of a dialog actively initiated by the man-machine dialog system is more pertinent and conforms to the progress of the current man-machine dialog, and user experience is improved.

(ZH)
本发明公开一种人机对话方法,应用于人机对话系统,所述方法包括:以人机对话系统的当前状态作为第一神经网络的输入,以确定人机对话系统对用户的主动对话方式;根据主动对话方式确定待推荐话题;以当前状态和待推荐话题的特征向量作为第二神经网络的输入来确定待推荐话题的推荐概率;根据推荐概率值从待推荐话题中选择待推荐知识点以呈现给用户。本实施例根据人机对话系统的当前状态来确定与用户之间的主动对话方式,从而能够保证人机对话系统所主动发起的对话形式是更具针对性的,符合当前人机对话的进展情况,提升了用户体验。