Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

A Neural Network Relation Classification Method and Its Realization System Fused with Distinguishing Degree Information

A technology of relationship classification and neural network, applied in the field of natural language processing, can solve problems such as easy confusion and different entity directions

Active Publication Date: 2019-11-08
SHANDONG UNIV
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] The present invention proposes a new feature (distinguishing degree information) to solve the problem that two types of relationships with the same relationship but different entity directions are easily confused

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 Neural Network Relation Classification Method and Its Realization System Fused with Distinguishing Degree Information
  • A Neural Network Relation Classification Method and Its Realization System Fused with Distinguishing Degree Information
  • A Neural Network Relation Classification Method and Its Realization System Fused with Distinguishing Degree Information

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0077] A neural network relationship classification method for fusing discrimination information, comprising the following steps:

[0078] (1) Data preprocessing: This application uses public data sets for result evaluation, but the public data sets are raw data, which do not meet the input requirements of the model and require preprocessing. First use the one-hot form to represent the entity words in the data set; for example, express the entities in the data set as 1.0, and the others as 0.0; then classify the data according to the text categories in the data set; the text categories in the data set are altogether It is divided into 19 categories, and a 19-dimensional one-hot vector is used to represent the category of each category. The position corresponding to 1 in the one-hot vector is the index position of the category. Put the 19-dimensional vector and the sentence in the text in the In the same line, separated by " / ", when reading the data, the samples and labels are ...

Embodiment 2

[0087] According to a kind of neural network relation classification method of fusing discrimination degree information described in embodiment 1, the difference is that,

[0088] In step (2), the training word vector includes:

[0089] A. Download the English data of Wikipedia for the whole day on November 6, 2011 as the initial training data, and clean these initial training data, remove meaningless special characters and formats, and process the data in HTML format into data in TXT format ;

[0090] B. Feed the data processed in step A into Word2vec for training. During training, use the skip-gram model, set the window size to 3-8, set the iteration cycle to 2-15, and set the dimension of the word vector to 200-400 Dimension, after training, get a word vector mapping table;

[0091] C. Obtain the word vector corresponding to each word in the training set according to the word vector mapping table obtained in step B. In order to speed up the training, this patent correspo...

Embodiment 3

[0132] The realization system of the above-mentioned neural network relation classification method, such as figure 1 As shown, it includes a sentence representation module, a discrimination module and a feature fusion module, and the sentence representation module and the discrimination module are respectively connected to the feature fusion module;

[0133] The sentence representation module is used to: correspond each word in the sentence in the training set to the dictionary, find its corresponding word vector, turn it into a vector form that can be processed by the computer, obtain the position vector, and combine the obtained position vector with the previous word vector Cascading, the obtained new vector is used as the input of the Bi-LSTM unit, and the semantic features of the sentence are obtained after being encoded by the Bi-LSTM unit;

[0134] The discrimination module is used to: Subtract the word vectors of the two entity words specified in the sentence, concatena...

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 neural network relation classification method and an implementation system thereof, which fuse discrimination degree information, comprising: (1) data preprocessing; (2) training word vector; (3) extracting position vector; Combining the word vector of each word with the high-dimensional position feature vector, the joint feature is obtained. (4) computing the semantic representation of sentences; The context information and semantic information of entities are encoded by bi-directional LSTM. (5) calculating a discrimination vector; After the two entity vectors are subtracted from each other and cascaded with a Bi-LSTM unit encode that input; (6) inputting the outputs of (4) and (5) to the CNN, outputting the feature vector fused with the discrimination information, and inputting the feature vector to the classifier for classification; (5) adopting a Loss function to train the model. The invention does not need to manually extract any features, and the jointmodel does not need to preprocess the data by means of other natural language processing tools, the algorithm is simple and clear, and the effect is best at present.

Description

technical field [0001] The invention relates to a neural network relation classification method and an implementation system for integrating discrimination degree information, and belongs to the technical field of natural language processing. Background technique [0002] With the advent of the intelligent age, the processing method of big data is developing in the direction of automation and intelligence, and various jobs are gradually replaced by intelligent machines. There are more and more intersections between human society and intelligent machines. In this era background , intelligent and convenient human-computer interaction is becoming more and more important. Therefore, the automatic construction technology of question answering system and knowledge base has received great attention and achieved some results in both industry and academia. These achievements are inseparable from the support of basic theories such as natural language processing, among which relation ...

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 Patents(China)
IPC IPC(8): G06F16/332G06F16/35
Inventor 李玉军王玥张文真
Owner SHANDONG UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com