Reply content generation method of dialogue robot and terminal device

A dialogue robot and content technology, applied in instruments, special data processing applications, electrical digital data processing, etc., can solve problems such as low diversity, sentence semantic loss, etc., to reduce overestimation, improve decoding strategy and loss function, and achieve good The effect of generalization ability

Active Publication Date: 2019-07-05
SHENZHEN GRADUATE SCHOOL TSINGHUA UNIV
View PDF6 Cites 25 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] In order to solve the problem of sentence semantic loss and low diversity caused by the reply generation of existing dialogue robots, the present invention provides a method for generating reply content of dialogue robots and a terminal device

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
  • Reply content generation method of dialogue robot and terminal device
  • Reply content generation method of dialogue robot and terminal device
  • Reply content generation method of dialogue robot and terminal device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0021] The embodiment of the present invention proposes a method for generating reply content of a dialogue robot for a dialogue system. First, collect the training dialogue samples as the neural network generation model, and prepare the data: obtain the dialogue text from the relevant dialogue platform, and perform data preprocessing, mainly including word segmentation, word frequency statistics, vocabulary construction, low-frequency word filtering, etc.; select a A neural network generation model based on the encoder-decoder structure is used as the basic network architecture; then, a word prediction network is introduced into the decoder of the selected neural network generation model, and the decoder is required to predict the current word in the target utterance in each step of decoding. Subsequences that have not yet been generated, and an additional loss function is added to the training process to optimize the word prediction network; in addition, the maximum entropy r...

Embodiment 2

[0069] In this embodiment, the existing English dialogue data set DailyDialog is used to divide the training set and the test set. There is no intersection between the training set and the test set. The model is trained on the training set, and the quality and quality of dialog generation are evaluated on the test set. diversity. DailyDialog is a multi-round dialogue dataset for daily chat scenes constructed by the publisher of the dataset by crawling spoken English dialogue websites. It contains dialogues in daily life, covers a lot of emotional information, and has many more natural dialogue patterns. Three dialogue generation models in the prior art were selected for comparison, as follows:

[0070] (1) Seq2Seq with attention mechanism (AttnSeq2Seq): The Seq2Seq model with attention mechanism has shown effectiveness in various natural language processing tasks.

[0071] (2) Hierarchical Encoder-Decoder (HRED): Since the multi-round dialogue history consists of a series of ...

Embodiment 3

[0078] Such as figure 2 As shown, it is a schematic diagram of a terminal device for generating a reply from a dialogue robot provided by an embodiment of the present invention. The terminal device for generating the reply of the dialogue robot in this embodiment includes: a processor, a memory, and a computer program stored in the memory and operable on the processor, such as a data processing program. When the processor executes the computer program, it implements the steps in the above embodiments of the method for generating reply content of each dialogue robot, for example figure 1 Steps S1-S6 are shown.

[0079] Exemplarily, the computer program may be divided into one or more units, and the one or more units are stored in the memory and executed by the processor to implement the present invention. The one or more units may be a series of computer program instruction segments capable of accomplishing specific functions, and the instruction segments are used to describ...

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 provides a reply content generation method of a dialogue robot and a terminal device. The method comprises the following steps of obtaining a dialogue text, and carrying out data preprocessing to obtain a training sample of a neural network generation model; selecting a neural network generation model based on an encoder-decoder structure; introducing a word prediction network into the decoder and adding a loss function into the word prediction network so as to correct an original negative log likelihood loss function; adding a maximum entropy regular term into the corrected original negative log likelihood loss function to obtain a final loss function; performing model training on the neural network generation model to obtain an optimal parameter; and using the trained neural network generation model to receive the input of a user and generate a corresponding reply. The method has the good generalization ability, and is not limited to the encoder-decoder model with a specific structure, and can be combined with any end-to-end model, so that the reply quality can be considered while the reply diversity is obviously improved, and the user has better interaction experience.

Description

technical field [0001] The invention relates to the field of computer natural language processing, in particular to a method for generating reply content of a dialogue robot and a terminal device. Background technique [0002] The reply content generation of the dialogue system belongs to the field of computer natural language processing. The dialogue system can be used in many fields, such as shopping assistants in shopping malls, voice assistants on mobile phones, etc., such as Microsoft Xiaobing and Siri, which we are familiar with. Applications of dialogue systems, dialogue systems have a wide range of potential applications and attractive commercial value. Non-task-oriented dialogue systems usually focus on talking to people in an open field, providing reasonable responses and entertainment functions, and have played a role in many practical applications. Statistics show that in online shopping scenarios, nearly 80% of the words are chat messages, and the way to deal w...

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): G06F16/332G06F17/27G06F17/22
CPCG06F40/12G06F40/30
Inventor 杨余久王艺如杨芳
Owner SHENZHEN GRADUATE SCHOOL TSINGHUA 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