Similar problem matching method based on Coarse2Fine network
A matching method and similarity matching technology, applied in biological neural network models, instruments, character and pattern recognition, etc., can solve the problem of long training and running time, achieve excellent matching effect, improve running speed and accuracy. Effect
Pending Publication Date: 2022-01-28
JIANGSU UNIV OF SCI & TECH
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
However, due to the huge size of the Bert and Erine models, training and running time-consuming, so the enhanced LSTM network, that is, the ESIM text reasoning network, is selected to implement similar problem matching technology.
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 moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0041] In order to make the purposes, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments These are some embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts shall fall within the protection scope of the present invention.
[0042] In order to be able to explain the present invention more clearly, the following explanations are given for the embodiments: the present invention gives a premise p and derives a hypothesis hypothesis h, and the goal of the loss function is to judge whether p is related to h, that is, whether it can be determined by p Derive h, therefo...
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
Login to View More Abstract
The invention discloses a similar problem matching method based on a Coarse2Fine network. The similar problem matching method comprises the following steps: 1, conducting training by using a data set to obtain a corresponding Coarse2Fine model; 2, performing coarse-grained similarity matching on to-be-processed statements and statements in a database by using a BM25 model in a Coarse2Fine model to obtain a plurality of matched statements; 3, extracting feature values of the to-be-processed statements and the matched statements in an ESIM network, and calculating difference values between the feature values of the to-be-processed statements and feature values of the plurality of matched statements one by one; and 4, extracting the matched statement with a small difference value from the ESIM network, and taking the matched statement as an output result of the similar statement. According to the invention, the problem that an optimal matching result cannot be selected due to too many high-score items caused by single use of ESIM network matching is solved.
Description
technical field [0001] The invention relates to the technical field of language matching, in particular to a similar problem matching method based on Coarse2Fine network. Background technique [0002] Similar question matching technology plays a key role in the current natural language processing applications such as intelligent customer service and chat robots. In the process of practical application, the core of similar problem matching is to establish a similar problem matching model. Common techniques are mainly divided into two types: one is the sentence vector representation based on word embedding stacking, and the similarity is judged by calculating the cosine vector between two sentences. The second is to calculate the semantics of two sentences, characterize the semantics of sentences through word embedding, RNN, LSTM, etc., and then use cosine to calculate distance or full connection and softmax layer to calculate the probability of sentence synonymy and non-synon...
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
Login to View More Patent Type & Authority Applications(China)
IPC IPC(8): G06F40/30G06K9/62G06N3/04
CPCG06F40/30G06N3/044G06F18/22G06F18/214G06F18/241G06F18/2415Y02D10/00
Inventor 王东升赵翠平王奇李佳伟路曼钟家国
Owner JIANGSU UNIV OF SCI & TECH