Word embedding deep learning method based on knowledge graph

Abstract

The invention discloses a word embedding deep learning method based on a knowledge graph. In the training sample set construction stage, firstly, entity relations in the knowledge graph are divided according to the semantic intensity, and then training samples with different path lengths are generated on the basis of the divided entity relation groups. In the word embedding deep learning stage, athree-task deep neural network structure composed of a word2vec encoder, a convolution neural network, a gating cycle unit network, a softmax classifier, a logic regression device and the like is constructed, and then parameters of the deep neural network structure are optimized by taking a training sample set generated in the previous stage as input through iteration. After training is completed,a word nesting encoder composed of the word2vec encoder and the convolution neural network is reserved. Compared with the prior art, the method has the advantage of being high in word embedding accuracy, high in generalization ability, easy to implement and the like, and can be effectively applied to the fields such as big data analysis, electronic commerce, intelligent traffic, medical health and intelligent manufacturing.

Description

Technical field [0001] The present invention relates to the field of computer application technology, in particular to a machine learning method for word embedding. Background technique [0002] Word embedding is a very important and widely used technology that can convert text and words into a one-dimensional numerical vector that the machine can accept. The length of the vector can be flexibly set according to needs. [0003] In the early days of word embedding technology, researchers proposed and used the one-hot method to convert words into one-dimensional vectors. The length of the vector is the size of the vocabulary, most of the elements are 0, and only one component has the value 1. This component represents the current word. For example, "Xiao Ming is a child star" is divided into "Xiao Ming|是|个|child star" after word segmentation. Then there are a total of four words in this sentence, so the code for "Xiaoming" is "0001", the code for "Yes" is "0010", the code for "one"...

AI Technical Summary

Problems solved by technology

[0006] We found that although the word2vec method can learn the one-dimensional vector representation of words to a certain extent, it is difficult to capture the deep feature representation of words because its training model only contains a shallow neuron network model with a hidden layer, and It does not consider the semantic information between the training data, so that the learned word vector lacks semantic association

Images