LSTM-ATTENTION TEXT CLASSIFICATION METHOD COMBINED WITH KEY INFORMATION, 1-7.

Jinbao Yang, Min Ma, Yu Fu, and Yanhong Gu

View Full Paper

Keywords

Text classification, points of text, attention mechanism, bidirectionallong shortterm memory (LSTM) network

Abstract

Aiming at the problem that the text information could not be fully understood in traditional text classification, this paper proposed a recurrent neural network classification model based on the attention mechanism combined with key information and gave a sentence points extraction method. The semantics of the text was first expressed by the sentence points, which could fully express the semantic information and enrich its feature, so as to solve the problem of the feature sparsity in the text classification. Then, the bidirectional long short-term memory network with the attention mechanism was selected as the classifier for learning. Set the word vector and sentence vector as the input of the network, respectively, the outcomes could be spliced to obtain the final category. Experimental results show that this method could improve the accuracy of text classification.

Important Links:



Go Back