Unlock instant, AI-driven research and patent intelligence for your innovation.

Personalized retrieval type dialogue method and system based on semantic representation

A technology of semantic representation and semantic matching, applied in the field of artificial intelligence, can solve the problems of poor correlation, small scale of personalized dialogue data sets, poor user experience, etc., and achieve the effect of easy acquisition, strong data scalability, and better acquisition.

Active Publication Date: 2021-01-05
RENMIN UNIVERSITY OF CHINA
View PDF7 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The method of using artificially constructed personalized dialogue datasets for research has the disadvantages of small scale of personalized dialogue datasets and poor data scalability
However, when using the method of extracting personalized features from user history, when the number of user feature sentences is large and the topics are relatively concentrated, the prediction results of the model may have a strong correlation with the feature sentences, but a poor correlation with the user input sentences, resulting in the model giving The reply is not a reasonable reply to the input sentence, resulting in poor reply from the model and poor user experience

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
  • Personalized retrieval type dialogue method and system based on semantic representation
  • Personalized retrieval type dialogue method and system based on semantic representation
  • Personalized retrieval type dialogue method and system based on semantic representation

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0026] This embodiment discloses a personalized retrieval dialog method based on semantic representation, such as figure 1 shown, including the following steps:

[0027] S1 extracts the personalized feature sentence p from the user's utterance history sentence set.

[0028] In this embodiment, the personalized features of the user are extracted through the speech history of the user on the social platform. All the sentences posted by the user on the social platform constitute the speech history of the user. These historical sentences contain a wealth of information, and some sentences can reflect the user's personality, hobbies and other characteristics. In this embodiment, the user's historical speech sentences are screened, and the sentences satisfying the following characteristics are used as personalized characteristic sentences p: 1) the length of the sentence is 5 to 30 characters; 2) the sentence contains "I"; 3) the sentence contains at least There is a verb; 4) ther...

Embodiment 2

[0053] Based on the same inventive concept, this embodiment discloses a personalized retrieval dialogue system based on semantic representation, including:

[0054] The feature extraction module is used to extract the personalized feature statement p from the user speech history statement set;

[0055] The candidate set module is used to search in the pre-established dialogue set according to the input sentence q given by the user combined with the personalized feature sentence p, and generate the candidate set C from the search results;

[0056] The sentence matching module is used to reply all candidate replies r={r in the candidate set C 1 , r 2 ,...,r n} match the user input sentence q and personalized feature sentence p to get each candidate reply r i Sentence matching score score(q, p, r i );

[0057] A reranking module for reranking candidate replies to r i The sentence matching scores are sorted, and the candidate reply with the highest score is selected as the f...

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 a personalized retrieval type dialogue method and system based on semantic representation. The personalized retrieval type dialogue method comprises the following steps that S1, extracting a personalized feature statement p from a user speaking historical statement set; S2, retrieving in a pre-established dialogue set according to an input statement q given by a user in combination with the personalized feature statement p, and generating a candidate set C according to a retrieval result; S3, performing statement matching on all candidate replies r={r1, r2,..., ri} inthe candidate set C, the user input statement q and the personalized feature statement p to obtain a statement matching score (q, p, ri) of each candidate reply ri; and S4, sorting the statement matching scores of the candidate replies ri, and selecting the candidate reply with the highest score as a final personalized reply. For an open domain dialogue system of a social media platform, personalized dialogue data is easier to obtain, and the data expandability is high.

Description

technical field [0001] The invention relates to a personalized retrieval dialogue method and system based on semantic representation, belonging to the technical field of artificial intelligence. Background technique [0002] The realization of personalized dialogue system can be mainly divided into two forms. One is to directly set the personalized features in the form of key-value pairs for the computer, that is, the specific age, gender, occupation and other information of the given system users. The system calculates and gives a reply corresponding to the speaking style of the user group according to this feature. The personalized dialogue system with fixed personality was realized through machine learning and rules in the early days. This kind of early personalized chat robot used the defined input and output mode, system personalization and system performance to determine the reply to a certain input. With the development of artificial intelligence, deep learning techn...

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): G06F16/33G06F16/332G06F40/211G06F40/216G06F40/30G06N3/04
CPCG06F16/3329G06F16/3344G06F40/30G06F40/211G06F40/216G06N3/049G06N3/044Y02D10/00
Inventor 窦志成马跃元文继荣
Owner RENMIN UNIVERSITY OF CHINA