Eureka AIR delivers breakthrough ideas for toughest innovation challenges, trusted by R&D personnel around the world.

A clustering pre-analysis-based multi-intention recognition and semantic slot filling combined modeling method

A recognition method and pre-analysis technology, applied in semantic analysis, neural learning methods, character and pattern recognition, etc., can solve the problem of higher level of semantic understanding, semantic information has not been effectively used, and the two aspects have not been fully considered. Task contact and other issues

Active Publication Date: 2021-08-03
NANJING UNIV OF POSTS & TELECOMM
View PDF1 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the traditional separate modeling method does not fully consider the connection between the two tasks, so that the semantic information cannot be effectively used.
In addition, human-computer dialogue systems often face the problem of multi-intent recognition, that is, the intention text entered by the user may not only contain one intention, but may also appear in multiple intentions.
At present, the research on intent recognition mainly focuses on the recognition of single intent. Compared with single intent recognition, multi-intent recognition is not only more complicated to recognize but also requires a higher degree of semantic understanding.

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
  • A clustering pre-analysis-based multi-intention recognition and semantic slot filling combined modeling method
  • A clustering pre-analysis-based multi-intention recognition and semantic slot filling combined modeling method
  • A clustering pre-analysis-based multi-intention recognition and semantic slot filling combined modeling method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0051] Such as figure 1 As shown, the present invention discloses a multi-intent and semantic slot joint recognition method based on clustering pre-analysis, including the following steps:

[0052] Step S101, obtaining the multi-intent text input by the current user in real time and performing preprocessing;

[0053] To preprocess the multi-intent text input by the current user is to vectorize the multi-intent text so that it can be input into the neural network model for semantic feature extraction. The vectorized representation method in this paper is to first use the massive Chinese unsupervised corpus in the same field to train the BERT model. Then, the obtained BERT pre-trained model is used to vectorize the multi-intent text.

[0054] Step S102, constructing a multi-intent recognition model based on clustering pre-analysis to identify multiple intentions of the user; figure 2 shown.

[0055] The purpose of building a multi-intent recognition model based on clusterin...

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 clustering pre-analysis-based multi-intention recognition and semantic slot filling combined modeling method. The method comprises the following steps: acquiring a multi-intention text input by a current user in real time and preprocessing the multi-intention text; constructing a multi-intention recognition model based on clustering pre-analysis, wherein the multi-intention recognition model is used for recognizing a plurality of intentions of the user; building a BiLSTM-CRF semantic slot filling model based on a Slot-Gated association gate mechanism, and guiding filling of semantic slots by fully utilizing an intention recognition result; and optimizing the constructed joint model of multi-intention recognition and semantic slot filling. According to the invention, the relation between intention recognition and semantic slot filling is fully considered, a joint modeling method is provided, two semantic analysis subtasks are combined into one semantic analysis task, the accuracy of semantic slot filling is improved while the accuracy of multi-intention recognition is improved, and therefore the quality of natural language semantic analysis is improved; in practical application, the human language understanding capability of a machine in man-machine conversation can be effectively improved, and the problem solving capability and the man-machine conversation experience are improved.

Description

technical field [0001] The invention relates to the field of natural language processing, in particular to a natural language semantic analysis method in a man-machine dialogue system. Background technique [0002] With the rapid development of artificial intelligence, people's requirements for the intelligence of many application scenarios are increasing. To meet the requirements of intelligence, good human-computer interaction is essential. At present, there are various ways of human-computer interaction, among which the most convenient way is to use natural language. Therefore, the voice of using self-speaking language to realize human-computer dialogue is getting higher and higher. The human-computer dialogue system has received extensive attention from academia and industry, and has a very wide range of application scenarios. [0003] To realize the human-computer dialogue system, the natural language semantic analysis technology is indispensable. The quality of seman...

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): G06F40/194G06F40/30G06K9/62G06N3/04G06N3/08
CPCG06F40/194G06F40/30G06N3/084G06N3/047G06N3/048G06F18/23213
Inventor 张晖李吉媛赵海涛孙雁飞朱洪波
Owner NANJING UNIV OF POSTS & TELECOMM
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
Eureka Blog
Learn More
PatSnap group products