Cached Long Short-Term Memory Neural Networks for Document-Level Sentiment Classification
Cached Long Short-Term Memory Neural Networks for Document-Level Sentiment Classification
Recently, neural networks have achieved great success on sentiment classification due to their ability to alleviate feature engineering.However, one of the remaining challenges is to model long texts in document-level sentiment classification under a recurrent architecture because of the deficiency of the memory unit.To address this problem, we present a …