A Text Classification Method Based on GCN

Abstract

The invention discloses a GCN-based text classification method, comprising: obtaining a text classification corpus; wherein, the corpus includes a plurality of samples, each sample includes a title and a chapter; preprocessing the corpus, and converting the preprocessed corpus The set is divided into training set, verification set and test set; the text is processed through spacy, and the graph relationship between words is extracted; according to the graph relationship, each word is embedded in the low-dimensional real-valued vector space of the matrix; according to the vector representation of the word, Construct a bidirectional LSTM and obtain a sentence representation; reconstruct the sentence representation based on the self-attention mechanism, input it into the GCN neural network, and train a semantic classification model; input the text word vector of the verification set into the model, and save the record on the verification set The model parameters when the effect is optimal; the optimal model based on the verification set is used to test the test set to obtain the classification result. The present invention uses LSTM and GCN plus an attention mechanism to finally obtain more accurate class results.

Description

technical field [0001] The present invention relates to the technical field of text classification, in particular to a text classification method based on GCN (Graph convolutional networks, graph convolutional neural network). Background technique [0002] In the past few years, with the rapid development of science and technology, especially the rapid development of the Internet and social networks, all kinds of information are flooding the Internet. Among them, the CSDN blog is developing rapidly, providing a platform for Internet technicians to develop and communicate. Experience, and share the solutions to the problems you encounter, etc.; for everyone to exchange and learn, and also to make a record of your own growth. With the development of the platform, the number of users has increased, and more and more articles have been published. People can obtain a large amount of information through the platform, but how to find the law from these information, discover curren...

AI Technical Summary

Problems solved by technology

[0004] The current text classification models have their own problems. For example, the fastText model does not consider the word order information in the network structure, and the TextCNN model cannot model longer sequence information when the word order information is considered.

Images