Open domain dialogue model and method for enhancing reply personalized expression

A model, personality technology, applied in the field of open domain dialogue generation

Active Publication Date: 2020-12-18
HUNAN UNIV
View PDF3 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] Existing technologies only focus on how to select personalized information to generate dialogues, and the selection of personalized information is very dependent on the personalization-related topics shown in the dialogue input, which makes it impossible for robots to communicate in many dialogue scenarios that lack personalized information. Proactively and fully leverage personalized information to generate personalized responses

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Open domain dialogue model and method for enhancing reply personalized expression
  • Open domain dialogue model and method for enhancing reply personalized expression
  • Open domain dialogue model and method for enhancing reply personalized expression

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0073] Attached below figure 1 The preferred embodiment of the present invention is further described, the present invention comprises a pair of sub-networks having the same encoder-decoder backbone, consisting of Context-Dominated Network (CDNet) and Persona-Dominated Network (PDNet) two sub-networks, CDNet is A dialogue generation model based on a memory network, which is mainly used to learn the ability to select a personality from a personalized file and ensure that the generated reply is semantically associated with the user input message; PDNet directly according to a pre-given Personality text tags generate replies, which are mainly used to learn the ability to fully express a personality in replies. The two sub-networks are trained alternately through multi-task learning. They update the parameters of the encoder-decoder backbone during the alternate training process, so that the entire model can obtain the personalized selection and personalization learned by the t...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention relates to an open domain dialogue model and method for enhancing reply personalized expression, and belongs to the technical field of open domain dialogue generation. The system comprises a pair of sub-networks with the same encoder decoder backbone, each sub-network is composed of a CDNet sub-network and a PDNet sub-network, the two sub-networks are alternately trained in a multi-task learning mode, parameters of the encoder decoder backbone are updated in the alternate training process, and the parameters of the encoder decoder backbone are updated in the multi-task learning mode. Therefore, the whole model obtains the capabilities of personalized selection and personalized embedding learned by the two sub-networks in the training process, the model alternately trains CDNet and PDNet in a multi-task training mode, the capabilities of the two sub-networks are learned, and personalized information is generated and replied more sufficiently. The method has the beneficialeffects that based on the personalized dual-network dialogue model, more personalized replies are generated in various dialogue scenes; and the leading effect of personalized information in the dialogue process is enhanced, and the personalized expression ability of the robot is enhanced.

Description

technical field [0001] The invention relates to an open domain dialog model and method for strengthening personalized expression of replies, and belongs to the technical field of open domain dialog generation. Background technique [0002] At present, dialogue systems can be roughly divided into two models, task-oriented and non-task-oriented, according to their specific applications. Task-oriented dialog systems are designed to help users complete certain tasks, such as finding products, booking accommodation and restaurants. Non-task-oriented dialogue systems are also known as open-domain dialogue systems or chat robots. They are dedicated to talking with people in the open domain and responding meaningfully and relevantly in the process of interacting with humans, mainly to shorten the distance between users and establish Trust relationship, emotional companionship, smooth dialogue process (for example, when task-based dialogue cannot meet user needs) and the role of imp...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06F16/332G06F16/33G06Q50/12G06N3/04
CPCG06F16/3329G06F16/3346G06Q50/12G06N3/045
Inventor 蒋斌周婉月杨超
Owner HUNAN UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap