Chinese text sentiment analysis using LSTM network based on L2 and Nadam
Jian Wang, Zewen Cao · 2017
The convenience of the network has led to the emergence of more and more commentary texts, most of which have user's opinion and experience, so mining opinions from these texts has become more and more important for many APPs and websites. However, such task is very challenging, in particular for Chinese review text. In this paper, we propose a text sentiment analysis method based on LSTM with L2 and Nadam optimizer to evaluate the accuracy of text sentiment analysis. The experiment results prove that the new optimization function and loss function improve the accuracy of the model and generalization ability and our LSTM based on L2 and Nadam model can get higher accuracy with fewer epochs.